Review of the Toxicology of Mineral Spirits

Abstract

This review of the toxicology of mineral spirits covers studies of the major classes of mineral spirits and several toxicologically important mineral spirit constituents. This review cites data from numerous previously unpublished animal toxicology studies conducted on mineral spirits during the past 30 years, expanding the existing database on the toxicology of this group of hydrocarbon solvents. The data can be used to better evaluate the potential effects associated with exposure to these materials, including health and environmental reviews such as the U.S. Environmental Protection Agency High Production Volume (HPV) chemical program and the Organization for Economic Cooperation and Development (OECD) HPV Screening Information Data Set (SIDS) program. The majority of animal toxicology studies in the available literature were conducted on mineral spirits categorized as ASTM D235 Type I Class A (149°C to 213°C boiling range; 8% to 22% aromatics) and demonstrate that Type I Class A mineral spirits have a low order of acute toxicity and do not produce significant systemic effects. Some additional studies conducted with ASTM D235 Type II Class C mineral spirits (177°C to 213°C boiling range; <2% aromatics) suggest that Type II Class C mineral spirits have similar toxicity to Type I Class A mineral spirits, though there is some evidence that Type II, Class C mineral spirits have a lesser degree of central nervous system (CNS) effects than the higher aromatic containing Type I Class A materials. In addition, toxicity data on selected chemical constituents of mineral spirits (e.g., n-nonane, n-decane, n-undecane) indicate that these chemicals have similar toxicological properties to mineral spirits. Overall, the data showed that mineral spirits have a low order of acute toxicity and do not appear to produce toxicologically relevant systemic effects. Ongoing studies are evaluating the concerns associated with chronic low-level exposure and central nervous system effects.

Mineral spirits is a generic term given to hydrocarbon solvents, which are complex substances consisting of numerous hydrocarbon components (paraffins, cycloparaffins, and aromatics) predominantly in the C₈ to C₁₃ carbon range and typically boiling in the range of 140°C to 220°C; other generic names for these substances include Stoddard solvent, white spirits, and high flash (140) aliphatic naphtha. The mineral spirits studies included in this review are on products that were derived from crude oil by distillation and have undergone a number of further refining steps, such as hydrodesulfurization and hydrotreating to remove sulfur and nitrogen compounds and to convert some or all of the aromatic components into saturated (aliphatic) species. Most of these mineral spirits also undergo further distillation to meet product specifications, such as vapor pressure and boiling range.

Although mineral spirits have higher molecular weights and a boiling range (140°C to 220°C) that is well above the boiling point for benzene (80°C), concern over benzene exposure from use of hydrocarbon solvents in general has resulted in these substances being closely controlled for benzene content for the past several decades. The specification for benzene content in mineral spirits is usually <0.1%, but in practice benzene levels are typically below 0.005% (<50 ppmv) due to the refining and distillation techniques. There have been several publications dealing with current and historical information on benzene levels in mineral spirits. Carpenter et al. (1975) reported that the sample of Stoddard solvent that was used in their toxicological studies contained 0.1% (1000 ppmv) benzene. Most subsequent reports on benzene levels in mineral spirits cite to Carpenter. In 1977, a study conducted by Battelle Columbus Laboratories on behalf of the U.S. Consumer Product Safety Commission (CPSC) reported that typical levels of benzene in mineral spirits were in the range of 0.01% to 0.03% (100 to 300 ppmv) (Hillman, 1978). More recently, McKee et al. (2007) noted that analytical methods optimized for the detection of low levels of benzene in complex hydrocarbon solvents were developed in the late 1970s. Since the late 1970s, levels of benzene in mineral spirits have been below approximately 50 ppmv (McKee et al. 2006), with specifically reported values of less than 20 ppmv (McKee et al. 2007; Fedorak et al. 2003).

Some of the Chemical Abstract Service (CAS) registry numbers and International Union of Pure and Applied Chemistry (IUPAC) names associated with the above materials are

In the American Society for Testing and Materials (ASTM) specification D235 (1995), four types of mineral spirits have been described. They are

Mineral spirit types may be further differentiated based on aromatics content as follows:

The nomenclature applied to mineral spirits described above has an interesting history. Until the 1960s, the predominant type of mineral spirits produced was a desulfurized straight-run (i.e., fractionated directly from crude oil by distillation) solvent, boiling in the range of 149°C to 204°C (300°F to 400°F), which was commonly known as Stoddard solvent, so named by the dry-cleaning industry. This material contained 15% to 25% aromatic compounds. With the advent of the United States Environmental Protection Agency’s (USEPA) Toxic Substances Control Act (TSCA), this product was entered in the TSCA inventory under CAS number 8052–41-3.

When clean air standards were first promulgated in the late 1960s and early 1970s, particularly in California (e.g., Rule 66 in the Los Angeles area, and Regulation 3 in San Francisco), lower aromaticity was desired and new mineral spirits grades were developed. In subsequent years, due to demands for various product performance and formulation needs, mineral spirits with much lower aromaticity (~<1%) have been produced. These de-aromatized mineral spirits are lower in odor than the higher aromatic versions and are popular in many applications, primarily coatings and consumer products, as well as in dry-cleaning. Hence, the name “Stoddard solvent” is now often applied to many types of mineral spirits even though these products are somewhat compositionally different from each other as well as from traditional Stoddard solvent. Composition and physical/chemical data on many commercial mineral spirits and their components are summarized in Table 1.

This review summarizes available toxicity data on mineral spirits and, to the extent possible, differentiates the toxicity by test material specification. Special emphasis was placed on identifying and summarizing all studies which have been conducted with products generally meeting ASTM Type I, Class A and ASTM Type II, Class C Mineral Spirits specifications. Products identified in this dossier, however, may not be tested to, or meet the standard specifications for mineral spirits outlined in ASTM D235. Studies conducted with other types and/or class of mineral spirits or components were also included when it seemed appropriate and justifiable.

MAMMALIAN TOXICOLOGY STUDIES

Acute Toxicity

Mineral spirits with up to 22% volume aromatics and typical components of mineral spirits, i.e., n-decane and n-undecane, have a low order of acute toxicity by the oral, dermal, and inhalation routes of exposure when tested in animals. Acute toxicity studies on mineral spirits and typical components are summarized in Table 2.

Based on the available data, the rat oral LD₅₀ for mineral spirits are greater than 5.0 g/kg (Shell 1964, 1976, 1977, 1979a; Hine and Zuidema 1970; ExxonMobil Biomedical Sciences, Inc. [EMBSI] 1975a, 1977a, 1977b; API 1986a). The dermal LD₅₀ values in rats and/or rabbits for mineral spirits are greater than 2.0 g/kg (Shell 1964, 1976, 1977, 1979a; Hine and Zuidema, 1970; EMBSI 1975a, 1984a, 1984b; API 1986a). The inhalation LC₅₀ (4- to 8-h exposures) in rats for mineral spirits is greater than 2.9 mg/L (Shell 1964, 1976, 1977, 1979a, 1988b; Hine and Zuidema 1970; Carpenter et al. 1975a, 1975b; EMBSI 1975a, 1975b, 1995; Riley et al. 1984; API 1987; Kulig 1989, 1990; Lammers 1997). It is important to note that testing was often limited by protocol design, responsible animal use provisions, and/or the physical/chemical properties of the tested materials and not by toxicity per se.

Mineral spirits also have a low order of acute toxicity in humans by the inhalation, dermal, and oral routes of exposure (IPCS 1996). However, as in the case of most hydrocarbon solvents with low viscosity, the severity of symptoms after ingestion of mineral spirits depends on whether the hydrocarbon solvent is aspirated into the lungs. Aspiration can cause serious bronchopneumonia. A dose of 30 ml aspirated into the lung may be fatal (McDermott 1975).

Irritation

Mineral spirits with up to 22% volume aromatics as well as components of mineral spirits, such as n-undecane and n-dodecane, can produce slight to severe dermal irritation in animals depending on the duration of exposure and exposure conditions (i.e., semioccluded, occluded patches, etc.) and the animal species used. They may also produce respiratory irritation in animals if administered at sufficiently high concentrations. However, they produce only minimal to negligible ocular irritation. Irritation studies on mineral spirits and their components are summarized in Table 3.

Primary dermal irritation studies in rabbits showed 4 hours exposure to mineral spirits containing less than 2 volume % aromatics resulted in slight to moderate skin irritation (Shell 1987, 1988a, 1988c, 1989a, 1989b, 1989c, 1993a, 1993b, 1995; EMBSI 1990). Mineral spirits containing up to 22 volume % aromatics showed similar results (EMBSI 1961a, 1986; Shell 1964; Hine and Zuidema 1970; Nethercott et al. 1980). However, exposures involving occlusive conditions and/or repeated exposures under conditions leading to defatting of the skin produced more severe skin responses (Shell 1976, 1977; API 1986a; Anderson et al. 1986; Jacobs et al. 1987; Ingram et al. 1993; Martinez et al. 1995).

In contrast, several animal studies have shown that mineral spirits containing up to 22 volume % aromatics produce minimal to negligible ocular irritation in rabbits (EMBSI 1961b, 1975a, 1979a, 1991a; Shell 1964, 1976, 1977, 1979a; Hine and Zuidema 1970; API 1986a).

Animal studies have also shown that mineral spirits produce no more than mild respiratory tract irritation at high concentrations (Shell 1990). Sensory irritation was evident as reduced breathing rates in test animals in several studies (Carpenter et al. 1975b; EMBSI 1988b), but at lower concentrations there was no evidence of sensory irritation for some materials (Carpenter et al. 1975a; EMBSI 1988a).

In humans, mineral spirits have been shown to produce slight to moderate skin irritation, and prolonged or repeated exposure, particularly when evaporation from the skin is prevented, can lead to severe irritant dermatitis. Nethercott et al. (1980) reported five cases of ulcerative and erythematous lesions of the genitals and the buttocks in workers wearing coveralls that had absorbed white spirit (Stoddard solvent) in dry-cleaning. An additional six cases of cutaneous irritation (vesicle formation, crusting, erythema, and desquamation) following skin contact with Stoddard solvent were also mentioned in this report. A human repeated-insult patch test with 100% C₉–C₁₃ mixed alkanes (n-, iso-, and cycloparaffins) under occluded patch resulted in severe dermal irritation (EMBSI 1991b). Due to the strong cutaneous reactions, the study was terminated prior to its scheduled end.

Slight eye, nose, and throat irritation has also been reported in humans following exposure to mineral spirits (Carpenter et al. 1975b; Hastings et al. 1984). Carpenter et al. (1975b) reported eye irritation and lacrimation in six volunteers after a 15-min exposure to Stoddard solvent at a vapor concentration of 2700 mg/m³. At 850 mg/m³, only one person reported slight eye irritation. No irritation was detected at 140 mg/m³. Hastings et al. (1984) found an increase in subjective reports of mild irritation symptoms during a 30-min exposure of 25 volunteers to a Stoddard solvent vapor concentration of 600 mg/m³. Irritation of the nose was experienced by 31% (15% in control group) and eye irritation by 36% (24% in the control group). No changes in the rates of eye-blinking, swallowing, or breathing were noted. In a more recent study, no exposure related changes in subjective observations were reported following exposure of 12 volunteers to either 57 or 570 mg/m³ white spirit for 4 h (Lammers et al. 2007). Astrand et al. (1975) found short-term trial exposures to 5000 mg/m³ white spirit produced subjective discomfort as nausea, vertigo, and a feeling of being severely affected that was present to a lesser degree at 2500 mg/m³. The symptoms were not present after 30-min exposure as high as 2500 mg/m³.

Sensitization

Mineral spirits containing up to 18 volume % aromatics have shown no evidence of skin sensitization. Skin sensitization studies on mineral spirits are summarized in Table 4.

The American Petroleum Institute (API 1986a) conducted a skin sensitization study with guinea pigs using the Buehler method and reported that Stoddard solvent containing 14.5% aromatics did not produce dermal sensitization. In this study, a 75% (v/v) solution of Stoddard solvent in a vehicle of paraffin oil was used for the three induction doses and a 25% (v/v) solution in paraffin oil was used as the challenge dose. Mild to moderate irritation was observed, but there was no evidence of sensitization. Similarly, a mineral spirit containing less than 5.0 volume % aromatics and one containing 18 volume % aromatics did not produce dermal sensitization when tested in the guinea pig maximization test (Shell 1976, 1977).

Repeated-Dose/Target Organ Toxicity

The most common effects observed in repeated-dose animal studies with mineral spirits are kidney changes (EMBSI 1981, 1991c; Phillips and Egan 1984; Phillips and Cockrell 1984; Shell 1979b, 1979c; Carpenter et al. 1975b; Viau et al. 1984; Doi et al. 2004). The initial gross observations confirmed that these changes occur in male but not female rats or in mice of either sex. Pathological observations include the presence of regenerative epithelium in the cortex and dilated tubules filled with proteinaceous casts in the corticomedullary areas. These kidney changes were described as “light hydrocarbon nephropathy” in the early publications, but further work on these and similar substances resulted in the identification of the underlying mechanism as an α _2u-globulin–mediated process (Doi et al. 2004) that, because of its sex and species specificity, is not regarded as relevant to humans (USEPA 1991). IARC (1998) identified a series of criteria to assess whether male rat kidney tumors are the consequence of an α _2u-globulin–mediated process. These include lack of genotoxicity, characteristic pathology, male rat specificity, α _2u-globulin accumulation, reversible binding of the α _2u-globulin to the test substance, in-creased cell proliferation, and similarities in dose-response relationships between histopathology and tumor outcome. The majority of these, including lack of genotoxicity, characteristic pathology, male rat specificity, α _2u-globulin accumulation, and increased cell proliferation, have been demonstrated for mineral spirits. A similarity of dose-response relationship between histopathologic changes in the kidney and kidney tumor outcome has not been specifically conducted, as no study has shown a statistically significant increase in renal tumors; however, the characteristic pathologic changes have been shown to occur at exposure levels that are not associated with excess kidney tumors (Doi et al. 2004). Finally, the aliphatic constituents of mineral spirit are similar to trimethyl pentane, a classic inducer of this effect (Swenberg et al. 1989).

Several studies also have reported effects in the liver, as well as hematological or urine chemistry changes, following repeated-dose exposure to mineral spirits (Jenkins et al. 1971; Shell 1979b, 1979c; EMBSI 1981, 1991c; Viau et al. 1984; API 1989). However, these effects have generally not been dose-related making their significance questionable. Repeated exposure toxicity studies on mineral spirits and their components are summarized in Table 5, and selected studies are discussed below.

Inhalation Toxicity

Less Than 2 Volume % Aromatics. Repeated-dose inhalation toxicity studies conducted on mineral spirits containing less than 2 volume % aromatics showed the only significant observation in rats was nephrotoxicity, consistent with the accumulation of α _2u-globulin. Phillips and Egan (1984) exposed groups of 35 male and female Sprague-Dawley rats to 300 or 900 ppm vapor of dearomatized mineral spirits containing less than 0.5 volume % aromatics for 6 h/day, 5 days/week for 12 weeks. No deaths occurred and only occasional decreased weight gain was noted in male animals exposed at the high levels. The male rat kidney was found to be the main target organ. After 4, 8, and 12 weeks of exposure, significant increases in absolute and relative kidney weights were found in all exposed groups, but were most apparent at the high exposure levels. Histopathological examination revealed the presence of regenerative epithelium in the cortex and dilated tubules filled with proteinaceous casts in the corticomedullary area of the kidney. The changes were focal in nature, covering 5% to 10% of the tubules. These observations were identical to the effects found when the authors reexamined the kidney slides from a study of Stoddard solvent by Carpenter et al. (1975b). Similar histological findings in male rats were also reported in an earlier 13-week study by Shell (1979c) at a dose of 259 ppm.

2 to 8 Volume % Aromatics. In a study by Carpenter et al. (1975a), rats and dogs were exposed to 140 Flash Aliphatic Solvent vapor containing 3% aromatics, at levels of 0, 0.05, 0.1, and 0.23 mg/L 6 h/day, 5 days/week for 72 days. No treatment-related effects on mortality, clinical signs, hematology, clinical chemistries, or histopathology were observed.

8 to 22 Volume % Aromatics. The majority of repeated-dose inhalation toxicity studies have been conducted with mineral spirits containing 8 to 22 volume % aromatics (Rector et al. 1966; Jenkins et al. 1971; Carpenter et al. 1975b; Shell 1979b; EMBSI 1979b; Phillips and Cockrell 1984; Riley et al. 1984).

In a study by Carpenter et al. (1975b), male Harlan-Wistar rats and dogs were exposed to Stoddard solvent vapor containing 15% aromatics, at levels of 0, 84, 190, and 330 ppm for 6 h/day, 5 days/week for 13 weeks. Histopathological lesions of the kidney and dilated tubules were found in 6 out of 9 rats exposed to 330 ppm and 3 out of 9 rats exposed to 190 ppm. These lesions were also noted in rats sacrificed after only 8 weeks of exposure. Significant changes in hematological values that were not dose related were thought to be mainly a consequence of unusual values found in the control group. There were no differences in body weight gain. No changes were observed in dogs with respect to body and organ weights, hematological and clinical chemical values, or histopathological parameters.

Rector et al. (1966) exposed rats, rabbits, dogs, and monkeys to concentrations of mineral spirits (13 to 19 volume % aromatics) ranging from 0.59 to 1.4 mg/L for 8 h/day, 5 days/week for 30 days, or to 0.11 to 1.27 mg/L mineral spirits continuously for 90 days. No treatment-related mortalities, body weight effects, hematology changes, or gross microscopic changes were seen in these studies at any exposure level. The only observed effects were some gross and microscopic changes indicative of lung irritation. When guinea pigs were exposed to the identical dosing regimen as stated above, effects were observed at 0.363 mg/L and above. Mortality, histopathological evidence of mild to moderate vacuolar changes in the liver, and elevated blood urea nitrogen levels were noted. However, the study in guinea pigs may have been confounded by a low vitamin C diet. In a subsequent study by Jenkins et al. (1971), to examine the apparent greater sensitivity of guinea pigs, mortality was much lower and liver changes were minimal in guinea pigs fed a diet high in vitamin C.

Shell (1979b) exposed rats to 0.0, 2.0, 4.0, or 7.5 mg/L (0, 345, 690, 1293 ppm) mineral spirit/white spirit product containing 18 volume % aromatics for 13 weeks. The major findings observed at all dose levels were increases in kidney weights and kidney lesions in male rats consistent with α _2u-globulin–mediated nephropathy. Low-grade anemia and increased spleen weight were also observed in males at all dose levels. At the highest dose, increased spleen weight was associated with minor histopathological changes. The significance of the anemia is uncertain. Increases in female liver weights were also observed, but there were no corresponding histopathological changes. Other small changes in serum chemistry parameters were not considered toxicologically important.

In another study, rats were exposed to 100 and 300 ppm of a C₈–C₁₃ mixed aliphatics and aromatics solvent containing 21 volume % aromatics for 6 h/day, 5 days/week for 12 weeks (EMBSI 1979b). There was no treatment-related mortality in this study. At week 12, the absolute kidney weights in the high-dose males and the relative kidney weights in both the low- and high-dose males were significantly elevated.

EMBSI (1979b) exposed Sprague-Dawley and Fischer 344 rats to 100 and 800 ppm of a C₈–C₁₃ mixed aliphatics and aromatics fluid containing 21 volume % aromatics for 6 h/day, 5 days/week for 8 weeks. There was no treatment-related mortality, and no adverse clinical signs or effects on hematology parameters were observed. Male body weights were slightly reduced in both strains at 800 ppm, but in Fischer rats returned to control values following a recovery period. Small changes observed in clinical chemistry values also returned to control values following a recovery period. Increased liver and kidney weights were observed in both sexes of both strains at study termination and remained increased in the Sprague-Dawley rats after a recovery period.

Oral Toxicity

Less than 2 Volume % Aromatics. A C₁₀–C₁₃ mixed alkanes (n-, iso-, and cycloparaffins) material, containing 0.5 volume % aromatics, was administered by gavage to rats at doses of 500, 2500, or 5000 mg/kg for 13 weeks (EMBSI 1991c). No treatment-related mortalities were observed. Male body weights were decreased in the 2500 and 5000 mg/kg dose groups during the latter stages of the study. This effect may be related to the stomach irritation noted during the pathological investigation. Also, during the latter stages of the study, an increase in food consumption was noted with increasing dose for both males and females, primarily in the 2500 and 5000 mg/kg dose groups. Liver weights were elevated in male rats at 2500 and 5000 mg/kg. Adrenal weights were significantly increased in male rats at 5000 mg/kg and in female rats at 2500 and 5000 mg/kg. Testis weights were elevated in male rats at 5000 mg/kg. Hyaline droplets and other changes (indicative of α _2u-globulin–mediated male rat nephropathy) were observed in male rat kidneys at all dose levels. Histopathological findings of hepatocellular hypertrophy were seen in livers of both sexes in all dose groups. Significant dose-related changes in hematology and serum chemistry parameters were observed and were consistent with changes seen in the liver. Other treatment-related effects were mucosal thickening and other signs of irritation of the stomach and anus that appeared to be the direct result of high dose intubation of a locally irritating substance. The treatment-related histopathological effects and corresponding organ weight increases were reversed during the 4-week recovery period and, thus, were considered adaptive in nature.

No subchronic oral studies were identified for mineral spirits with 2 to 22 volume % aromatics.

Dermal Toxicity

8 to 22 Volume % Aromatics. API (1989) conducted a repeated-dose dermal toxicity study in the rabbit on a Stoddard solvent containing 14.5 volume % of aromatics. Doses of 200, 1000, and 2000 mg/kg were applied for 6 h per day, 3 times per week for 4 weeks under occluded exposure conditions. At 200 and 1000 mg/kg, marked to severe skin erythema and edema were observed with histopathological changes to the skin. In the 1000 mg/kg dose group, reduced body weight gain was observed in females. In the 2000 mg/kg dose group, severe irritation and reduced body weight gain were observed in both sexes, and white granular foci in the liver were observed in females.

Human Epidemiological Studies

Several human epidemiological studies and case reports have evaluated the effects of occupational exposure to paints and mixed solvents. These studies and the results are discussed in Epidemiological Studies.

Nervous System Effects

A number of acute and subchronic inhalation neurotoxicity studies were conducted in rats exposed to white spirits (Savolainen and Pfaffli 1982; Kulig 1989, 1990; Lam et al. 1992, 1994, 1995; Ostergaard 1993; Bondy et al. 1995; Lund et al. 1996; Steensgard et al. 1996; Haas et al. 2001; Lammers et al. 2007; Hissink et al.. 2007). In these studies, numerous end points were assessed including, but not limited to visual discrimination performance, coordinated movement, spontaneous activity, grip strength, and peripheral nerve conduction time. In some subchronic studies, food and water intake, neuropathology evaluations, and/or reversibility of effects were also evaluated. Studies of the effects of mineral spirits and their components on the nervous system are summarized in Table 6, and selected studies are discussed below.

Nervous System Effects Following Inhalation Exposure

Less than 2 Volume % Aromatics. Hass et al. (2001) investigated the potential developmental neurotoxicity of dearomatized white spirit containing less than 0.4 volume % aromatics. In this study, rats were exposed to 0 or 800 ppm (4679 mg/m³) dearomatized white spirit for 6 h per day on gestation days 7 to 20. Developmental and neurobehavioral effects in the offspring were investigated using a test battery including assessment of physical development, reflex ontogeny, motor function, motor activity, and learning and memory. No significant effects were recorded on motor function or on activity in the open field test. In the initial learning period (age 1 month), the performance in a Morris water maze was similar in exposed and control animals. When testing for memory at the age of 2 months, the exposed male offspring used more time to locate the hidden platform. After platform relocation, impaired cognitive function was revealed in the exposed females. At the age of 5 months, learning and memory deficits were observed in exposed offspring. The differences were not related to poorer swimming capabilities, because swim speeds were similar to control values. Thus, the authors concluded that prenatal exposure to 800 ppm white spirit produced long-lasting learning and memory deficits in rats. A limitation of this study is that only one experimental dose, i.e., 800 ppm, was tested. It was not possible to assess the dose-response relationships and/or to determine a no-observed-adverse-effect level.

Lund et al. (1996) investigated whether exposure to dearomatized white spirits containing 0.4 volume % aromatics produces long-lasting morphological, electrophysiological, or neurobehavioral changes in rats. In this study, rats (3 months old) were exposed to dearomatized white spirit (0, 400, or 800 ppm) for 6 h/day, 5 days/week for 6 months. After an exposure-free period of 2 to 6 months duration, neurophysiological, neurobehavioral, and macroscopic pathologic examinations were performed. The authors reported exposure-related changes in sensory-evoked potentials and a decrease in motor activity during dark (no light) periods. However, there were no test substance–related changes in learning and memory functions. As the changes in sensory evoked potentials and motor activity were persistent, Lund et al. concluded that a 6-month exposure to 400 or 800 ppm dearomatized white spirit could produce long-lasting and possibly irreversible effects in the nervous system of the rat.

8 to 22 Volume % Aromatics. Kulig (1989) conducted both acute and subchronic studies with vapors containing 20 volume % aromatics. Male Wistar rats were exposed to 0, 200, 400, or 800 ppm white spirit vapors for 8 h/day for 3 consecutive days with behavioral tests conducted immediately after exposures (Kulig 1989). This study confirmed that acute white spirit exposure at 200 to 800 ppm could produce transient behavioral effects. In a subchronic study, male Wistar rats were exposed to 0, 200, 400, or 800 ppm white spirit vapors for 8 hrs/day, 5 days/week for 26 weeks, with behavioral tests generally conducted weekly at least 10 h after the last daily exposure. No persistent changes in neurobehavioral functioning were observed. In addition, no exposure-related changes in brain, spinal cord, or sciatic nerve were seen in light microscopy studies. These findings are consistent with earlier subchronic studies conducted by Carpenter et al. (1975b) on Stoddard solvent and by Phillips and Egan (1984) on dearomatized white spirit. In these earlier studies, no histopathological abnormalities of the nervous system were observed.

Ostergaard et al. (1993) examined the behavioral effects of exposure to white spirits containing 20 volume % aromatics in adult (3-month) and aged (15-month) rats. In these experiments, rats were exposed to white spirits (0, 400, or 800 ppm) for 6 h/day, 5 days/week for 6 months. After an exposure-free period of 2 months, neurobehavioral, pathological, and neurochemical examinations were performed. There were no white spirit-induced neurobehavioral effects. As expected, age-related differences in motor activity were detected; however, no dose-related macroscopic or histopathological changes were found. The concentrations of neurotransmitters noradrenaline (NA), dopamine (DA), and 5-hydroxytryptamine (5-HT) in various brain regions and in whole brain were changed in the 400 and 800 ppm groups (see also Lam et al. 1995). The authors concluded that this study revealed irreversible changes within the central nervous system (CNS). However, the significance of changes in neurotransmitter levels is difficult to evaluate particularly in the absence of any pathological or functional changes.

Nervous System Effects Following Dermal Exposure

Several studies were conducted to assess changes in neurotoxicity endpoints in rats following dermal exposure to white spirits containing up to 17 volume % aromatics (Verkkala et al. 1984). White spirits (210 to 260 mg) were applied to the tail 3 h/day, 5 days/week for 6 weeks. No change in motor conduction velocity was observed, although some electrophysiological changes occurred. Morphological analysis of the tail nerve revealed axonal swelling, widening of the nodes of Ranvier, and/or demyelinated foci in the axons.

Other Studies

Due to growing concerns regarding the possible neurotoxicity of mineral spirits and limitations of existing data, a neurobehavioral test program was conducted at TNO Laboratories in the Netherlands to develop a database on acute hydrocarbon solvent neurotoxicity and provide additional data for setting occupational exposure limits for hydrocarbon solvents. Neurobehavioral methods that can be used to examine the effects of acute exposures to hydrocarbon solvents in the C₅–C₁₁₊ carbon range were evaluated. In these studies, twelve hydrocarbons or hydrocarbon mixtures, with carbon chain lengths ranging from C₅–C₁₁, were tested for acute neurotoxicity. Male WAG/Rij/CrlBR rats were exposed by inhalation for 8 h on 3 consecutive days and evaluated daily for changes in motor activity, functional observation measures, and learned performance of a visual discrimination task. C₅ aliphatics (pentane, isopentane, cyclopentane) did not cause significant effects at exposures up to 20 g/m³, and C₇/C₈ aliphatics (methylcyclohexane, n-octane, isooctane) did not cause any significant effects at exposures up to 14 g/m³. n-Decane caused some limited, but statistically significant neurobehavioral effects at 5 g/m³, with no effects at 1.5 g/m³. A commercial solvent containing a mixture of C₉ aromatics caused significant neurobehavioral effects at 5 g/m³ and reversible effects on learned performance at 1 g/m³, with a no-observed-adverse-effect level (NOAEL) of 0.2 g/m³. The acute neurobehavioral data developed from this test program supplements existing toxicological data (Nessel et al. 2000; McKee et al. 2005). In addition to these studies in rodents, studies were also conducted in volunteers. Following exposure to mineral spirits, human subjects were given a questionnaire to assess their subjective symptoms. Computerized tests of learning and memory tasks were performed and blood levels of specific hydrocarbon markers (n-decane and trimethyl benzene) were also measured (Hissink et al. 2007; Lammers et al. 2007).

In the neurobehavioral test program, a series of studies were conducted to determine whether exposure levels associated with acute central nervous system (CNS) effects of white spirit (WS) in humans could be predicted from toxicology data in animals through the use of physiologically based pharmacokinetic (PBPK) models (Hissink et al. 2007). Experimental studies were conducted to measure acute CNS effects and toxicokinetic parameters in rats and human volunteers. WS was used as the model complex hydrocarbon solvent and two marker compounds, 1,2,4-trimethyl benzene (TMB) and n-decane (NDEC), were analyzed to characterize internal exposure after WS inhalation. Toxicokinetic data for blood and brain concentrations of the two marker compounds in the rat, together with in vitro partition coefficients, were used to develop PBPK models for both TMB and NDEC. The rat models were then allometrically scaled to obtain models for inhalation exposure for man. The human models were validated with blood and alveolar air kinetics of TMB and NDEC, measured in human volunteers. In general, the curves predicting the blood and brain levels in rats and the blood levels in humans fit well with the measured data.

Hissink et al. (2007) hypothesized that similar brain concentrations of TMB and NDEC would equate to similar effect levels in rats and humans. Using the PBPK model to test the hypothesis, human exposure effect and no effect levels were predicted from the CNS effect data and the measured TMB and NDEC brain concentrations in rats. Thus, a WS exposure concentration of 344 mg/m³ was predicted to result in a human brain concentration of TMB equivalent to that in the brains of rats exposed to 600 mg/m³ WS. A similar calculation for NDEC in human brains equated to an external exposure of 721 mg/m³ WS. From these results the no effect level for acute CNS effects in humans was predicted to be in the range of 344 to 721 mg/m³. As a test of that prediction, volunteers were exposed to WS for 4 h at approximately 570 mg/m³, and a number of neurobehavioral parameters were monitored. Of these, the only statistically significant finding was a small change in reaction time. Thus, it was determined that 570 mg/m³ represented either a no effect or a minimal effect level for acute CNS effects in humans. Of note, TMB was found to be the compound with the lower estimated no-observed-effect level (NOEL) and lowest-observed-effect level (LOEL). This indicates that aromatic compounds might be more important than aliphatics, for the acute CNS effect of WS.

In summary, the results of available animal studies on mineral spirits containing up to 22 volume % aromatics demonstrate that the acute CNS effects produced by concentrations as high as 800 ppm are completely reversible. The potential for persistent effects is less clear. Several studies have reported changes in behavior, neurochemistry, and sensory evoked potentials that the authors have interpreted as evidence of long-lasting, perhaps irreversible, changes. However, these findings have generally been subtle and not obviously associated with functional deficits or behavioral changes or pathological changes. Therefore, it is not clear whether these finding are a reflection of a true neurotoxic effect or simply differences from baseline values without toxicological significance.

Reproductive/Developmental Toxicity

Animal Studies

A number of animal studies have been conducted to evaluate the potential reproductive and/or developmental effects of mineral spirits. These data, although limited, suggest that mineral spirits with up to 24% volume aromatics and major components of mineral spirits (i.e., decane and undecane) are not likely to be reproductive or developmental toxicants. Reproductive and developmental toxicity studies on mineral spirits and their components are summarized in Table 7.

Less than 2 Volume % Aromatics. A C₉–C₁₃ mixed alkane (n-, iso-, and cycloparaffins) material containing 0.4 volume % aromatics was administered to rats by inhalation (300 and 900 ppm) for 6 h/day on days 6 to 15 of gestation (EMBSI 1978b). There were no treatment-related effects on pregnancy rate, mortality, body weight gain, and gross postmortem observations. There were no effects on uterine parameters (mean number of corpora lutea, implantations, or resorptions), number of viable fetuses, fetal size, or sex distribution data. In addition, the incidence of fetuses with ossification variations and external or visceral malformations in the treated groups were comparable to the negative control.

8 to 22 Volume % Aromatics. Jakobsen et al. (1986) conducted two experiments to evaluate the effects of white spirits (17.0 volume % aromatics) on pregnant Wistar rats. Rats were exposed to white spirits (0, 273, 483, or 953 ppm) for 6 h/day on days 6 to 15 of gestation, or 950 ppm white spirits on days 3 to 20 of gestation. Maternal toxicity was observed at 953 and 950 ppm. However, there were no differences in number of fetuses, and there were no increases in skeletal or visceral anomalies or malformations. In the 950 ppm group, mean fetal body weight was 14% lower than controls. In addition, the incidence of signs of delayed ossification and the number of fetuses with extra ribs was significantly increased. These changes were attributed to maternal toxicity. No malformations were seen.

A C₈–C₁₃ mixed aliphatics and aromatics material containing 21.0 volume % aromatics was administered to rats by inhalation (100 or 300 ppm) for 6 h/day on days 6–15 of gestation (EMBSI 1979c; Phillips and Egan 1981). No treatment-related differences were found with respect to implantation, number of live fetuses, fetal resorption, fetal size, or sex distribution. Furthermore, no external or visceral malformations were observed.

In another study using Stoddard solvent containing 24 volume % aromatics, pregnant rats were exposed to 0, 100, and 400 ppm for 6 h/day on days 6 to 15 of gestation (API 1977). No maternal toxicity was observed, and there were no differences in litter size or mean fetal weight between the groups. An increased incidence of fetuses with skeletal variations was observed in the exposed groups. However, these variations were not dose-related and there was no increased incidence of any type of malformation.

Maraschin et al. (1995) administered 0, 25, 150, and 1000 mg/kg/day n-decane to rats by oral gavage. Male rats were treated from day 14 prior to mating until the end of the mating phase. Female rats were treated from day 14 prior to mating through day 4 of lactation, and they were then sacrificed on day 5 of lactation along with their pups. The developmental toxicity assessment included observations of external abnormalities, number of live and stillbirths, mortality, sex determination, and weights of pups. In this study, no statistically significant treatment-related effects were observed at any dose level in any of the reproductive or developmental parameters. Thus, the NOAEL for both reproductive and developmental toxicity was 1000 mg/kg/day, the highest dose tested.

The Ministry of Health and Welfare Japan (1996) administered 0, 100, 300, and 1000 mg/kg/day n-undecane to rats by oral gavage. Male rats were dosed for 46 days including 14 days prior to mating, and during the mating period. Female rats were dosed for 14 days prior to mating, during the mating and gestation periods and postnatally until the third day of nursing. No treatment-related effects were observed at any dose level in any of the reproductive parameters evaluated in this study. However, body weight gain was decreased in male and female offspring at the 1000 mg/kg/day dose level. No effects were noted in terms of viability or in the general condition of the offspring. Thus, in this study, the NOEL for both reproductive and developmental toxicity was 300 mg/kg/day.

Human Studies

There are no studies on the reproductive or developmental effects of white spirits in humans. There are reports in the literature that attempted to link adverse effects on human reproductive functions with solvent exposure. However, no studies could be identified from which the potential effects of white spirits could be adequately assessed or evaluated. These studies are discussed in Epidemiological Studies.

Genotoxicity

In Vitro and In Vivo Studies

A review of the in vivo and in vitro studies on mineral spirits containing up to 22 volume % aromatics and components of mineral spirits suggest that these materials are not genotoxic (Table 8).

In Vitro Studies. Mineral spirits ranging in aromatic content from 0% to 19% and components of mineral spirits, including n-nonane, n-decane, and n-dodecane, did not produce an increase in mutations when tested in the Ames Salmonella typhimurium assay in either the presence or absence of metabolic activation (aAPI undated, 1978a, 1978b; Gochet et al. 1984; EMBSI 1991e; Shell 1998a; Zeiger et al. 1992; PETRESA 1985; Tummey et al. 1992). Similarly, negative results were obtained when Stoddard solvent (19 volume % aromatics) was tested in a mutagenesis assay with Saccharomyces cerevisiae (API 1978a).

Stoddard solvent (19% aromatics) was not genotoxic in a mouse lymphoma mutation assay in either the presence or absence of metabolic activation (API 1978a). Although another Stoddard solvent (14.5 volume % aromatics) later tested positive in a second mouse lymphoma assay (in the presence and absence of metabolic activation), the results of this study were equivocal because the doses used were cytotoxic (API 1986d).

The ability of white spirits (15 volume % aromatics) to induce sister-chromatid exchange (SCE) in mammalian cells in vitro (human peripheral blood cell cultures) was evaluated by Gochet et al. (1984). No SCEs were produced.

In Vivo Studies. Stoddard solvent (19 volume % aromatics) was tested in an in vivo rat bone marrow cytogenetics assay (API 1978a). In this assay, the rats were injected intraperitoneally with Stoddard solvent. No chromosomal aberrations were observed in the bone marrow cells.

Micronucleus tests were conducted in mice with white spirits (15 volume % aromatics) and a C₁₀–C₁₃ mixed alkane (n-, iso-, and cycloparaffins) material (0.5 volume % aromatics). Micronuclei were not observed in either study (Gochet et al. 1984; EMBSI 1991d).

Dominant lethal studies were conducted in mice and rats with Stoddard solvent containing 14.5 volume % aromatics and 140 Flash Aliphatic Solvent containing 3.0 volume % aromatics (API 1973). Lethal mutations were not produced in any study. In addition, dominant lethal mutations were not produced in rats treated with a C₉–C₁₃ mixed alkane (n-, iso-, and cycloparaffins) material containing 0.4 volume % aromatics, or a C₈–C₁₃ mixed aliphatic and aromatic material containing 21 volume % aromatics (EMBSI 1978c, 1980; Phillips and Egan 1981).

Human Studies

No studies have been conducted to determine the potential genotoxic effects of white spirit on humans. However, in vitro and in vivo data from animal studies suggests that mineral spirits are unlikely to be genotoxic.

Carcinogenicity

The National Toxicology Program (NTP) conducted 2-year carcinogenicity studies in rats and mice with Stoddard Solvent IIC (NTP 2004; Doi et al. 2004; Table 9). In these studies, F344/N rats (50 males and 50 females/group) were exposed to Stoddard solvent IIC by inhalation at concentrations of 0, 138 (males), 550, 1100, or 2200 (females) mg/m³, 6 h per day, 5 days per week for 105 weeks (NTP 2004). Additional groups of 10 males and 10 females were exposed to the same concentrations for 3 months for renal toxicity analyses. Survival in the top exposure concentration groups of males and females was significantly less than that of the chamber controls. Mean body weights of exposed males and females were similar to those of the chamber controls.

Cell proliferation analyses were performed in the left kidney of males and females after 3 months of exposure. The mean numbers of labeled cells and the labeling indices in males exposed to 550 and 1100 mg/m³ were significantly increased. The amount of α _2u-globulin in the right kidney of males increased with increasing exposure concentration. Also, the incidences of granular casts and cortical tubule degeneration and regeneration were generally increased in exposed males, as was the severity of hyalin droplets. These changes were not observed in females.

At 2 years, the incidences of benign and benign or malignant pheochromocytoma (combined) of the adrenal medulla occurred with positive trends in males, and the incidences in the 550 and 1100 mg/m³ groups were significantly increased. Due to an increased incidence of renal tubule hyperplasia in males at 2 years, extended kidney evaluations were conducted. A slightly (but not statistically significant) increased incidence of renal tubule adenoma occurred in the 1100 mg/m³ exposure group. Non-neoplastic lesions of the kidney characteristic of an α _2u-globulin accumulation occurred in the kidney of males exposed to Stoddard solvent IIC after 9 and 90 days of exposure.

B6C3F₁ mice (50 males and 50 females/group) were exposed to Stoddard Solvent IIC by inhalation at concentrations of 0, 550, 1100, or 2200 mg/m³, 6 h per day, 5 days per week for 105 weeks (NTP 2004). Survival of exposed mice was similar to that of the chamber controls. Mean body weights of exposed females were greater than those of the chamber controls. The incidences of hepatocellular adenoma occurred with a positive trend in females, and the incidence of multiple hepatocellular adenoma in females exposed to 2200 mg/m³ was significantly increased. However, the incidences of hepatocellular adenoma or carcinoma (combined) and hepatocellular carcinoma alone in exposed males and females were not significantly increased.

Based on the data from the 2-year carcinogenicity studies described above, the NTP concluded that there was some evidence of carcinogenic activity of Stoddard Solvent IIC in male F344/N rats based on the increased incidence of adrenal medulla neoplasms; the slightly increased incidences of renal tubule adenoma may have been related to Stoddard solvent IIC exposure. There was no evidence of carcinogenic activity of Stoddard solvent IIC in female F344/rats exposed to 550, 1100, or 2200 mg/m³. There was no evidence of carcinogenic activity of Stoddard Solvent IIC in male B6C3F₁ mice exposed to 550, 1100, or 2200 mg/m³. There was equivocal evidence of carcinogenic activity of Stoddard solvent IIC in female B6C3F₁ mice based on an increased incidence of hepatocellular adenoma; this slight increase was associated with increased body weight in exposed females.

As discussed above, an increased incidence of total pheochromocytomas as well as severe nephropathy was observed only in male rats at the highest exposure group. The possible correlation between the severity of chronic progressive glomerulonephropathy (CPN) and the incidence of adrenal pheochromocytomas was examined by Nyska et al. (1999) in studies of male Fischer 344 rats by the NTP. The NTP historical control database was first examined in order to determine whether there was an association between the severity of CPN and the occurrence of adrenal pheochromocytomas in unexposed animals. This analysis showed that pheochromocytomas were consistently observed in control animals with severe nephropathy, indicating that control animals with high severity grade nephropathy were at increased risk of developing pheochromocytomas. Following this analysis, the 125 most recent NTP studies conducted in F344 rats were examined to determine how frequently chemicals that cause increased severity of chronic nephropathy showed an increased incidence of pheochromocytomas. In contrast to the results with control animals, this evaluation showed that chemically related increases in nephropathy severity were generally not associated with pheochromocytomas. Because Stoddard solvent IIC exposure resulted in increases in both nephropathy severity and pheochromocytoma incidence, the correlation of these findings was also tested by Nyska et al. (1999). The resulting correlation was weak and not statistically significant, suggesting that the exposure-related increase in pheochromocytomas in male rats was unrelated to the severity of nephropathy.

In summary, the mechanism by which Stoddard Solvent IIC induces adrenal medullary neoplasms remains unclear. However, it must be emphasized that no genotoxic potential was identified by in vivo and in vitro genetic toxicity tests with Stoddard Solvent IIC, in the absence or presence of metabolic activation (NTP 2004). These results suggest that Stoddard Solvent IIC components are not metabolized to genotoxic compounds, an indication of low carcinogenic potential. The absence of increased tumor incidence in organs other than the adrenal medulla is consistent with these assumptions.

EPIDEMIOLOGICAL STUDIES

There are many epidemiological studies that have investigated general solvent exposure, but relatively few have evaluated exposure to mineral spirits specifically. This review addresses only published studies that have evaluated exposure to mineral spirits separately, or have identified mineral spirits as the predominant exposure. Recent studies of painters are considered because painters have been recognized as having potentially high exposures to this solvent category (IPCS 1996). However, no case studies or proportional mortality studies have been included in this review because of the serious limitations in their designs. The weight of the evidence suggested that long-term human exposure to hydrocarbon solvents at or below the occupational exposure limits (OELs) does not result in adverse neurobehavioral effects or any adverse birth outcomes. Several associations with cancer have been reported, but these findings are generally weak, inconsistent as to cancer site, and susceptible to bias. The epidemiology studies that were reviewed are summarized in Table 10.

Limitations Common to Many Mineral Spirits Epidemiology Studies

To adequately interpret and evaluate the epidemiological studies conducted on mineral spirits, the inherent limitations that these studies share with other observational research must be considered. These studies are not randomized experiments and are therefore subject to biases that may cause spurious results. Potential limitations are discussed below.

Recall Bias. Many of the evaluated studies were subject to recall bias because they relied on the participants’ subjective recall of exposure and/or disease (e.g., miscarriage). This may have resulted in individuals with disease/exposure being able to recall exposure/disease more frequently, thereby artificially elevating risks.

Confounding. Although this current review focused on exposures specific to mineral spirits, subjects had a considerable degree of coexposure to multiple environmental agents. This makes it very difficult to distinguish the independent effects of individual agents. Similarly, there are other associated risk factors for disease that may confound the true exposure-disease relationship. These risk factors were often not controlled or incompletely controlled in these studies, resulting in some degree of residual confounding. However, residual confounding is a less probable explanation for those study results that demonstrated moderate to strong excess risks (i.e., significant risk ratios that are approximately ≥3.0).

Misclassification Bias. Exposures in the reviewed studies were assigned relatively crudely, based on subjective recall, membership in a trade union, or occupation listed in a registry. None of the studies used more sophisticated methods that are available in exposure assessment, such as industrial hygiene measurements, biomarkers, or job exposure matrices. The magnitude and direction of the misclassification bias introduced by the crude type of exposure assessment cannot be easily estimated. It has been shown that even apparently nondifferential misclassification can cause spurious positive or negative results (Dosemeci et al. 1990; Flegal et al. 1991; Birkett 1992; Sorahan and Gilthorpe 1994; Wacholder 1995).

Multiple Comparisons. Most of the reviewed studies were exploratory in nature, typically evaluating multiple exposures, exposure combinations, and/or diseases. The effect of multiple comparisons is to increase the significance level threshold, or change a significant p < .05 to a smaller p value if statistical significance is relevant. An example of multiple testing is Siemiatycki et al. (1987a), who described an initial screening process that evaluated hundreds of cancer/exposure associations, followed by many additional in-depth analyses. Many of the statistically significant results arising from these multiple comparisons could be chance events that are not indicative of true associations between exposure and disease.

Publication Bias. It is important to remember that the published literature does not represent a comprehensive listing or random sampling of all research findings. Results that demonstrate links between exposure and disease are much more likely to be published, biasing this body of literature toward positive findings. This bias also provides a strong incentive for investigators to search their data carefully for positive results that can enhance chances for publication. Publication bias is seen as a serious problem even for randomized drug trials, which require formal protocols and strict Federal regulation (Simes 1986; Dickersin et al. 1987; Dickersin 1990; Easterbrook et al. 1991). Epidemiological studies are often more informal and could be subject to even greater bias toward positive associations.

Target Organ Toxicity Studies

Ship engineers commonly use white spirits and fuel oil (diesel oil) for washing and degreasing purposes in the engine room and to wash their hands. Svendsen and Hilt (1997) compared prevalence of skin disorders (acne, dry skin, any dermatitis, hand dermatitis) of 169 current engineers to 295 seamen who had never worked as ship’s engineers. Engineers had a higher prevalence for all conditions that were significant for dermatitis and dry skin on the hands.

Three case-control studies investigated the impact of solvents on chronic disease. McNamee et al. (1994) evaluated the occupational exposure histories of 102 cases of chronic pancreatitis and 205 controls and found an inverse trend with exposure to “paint solvents” (odds ratio [OR] = 0.87 at high exposure) after conditional logistic adjustment for alcohol use, diet, social class, and smoking. Lacey et al. (1999) reported significantly increased risks for undifferentiated connective tissue disease (UCTD) among those exposed by both self-report and expert review to “mineral spirits, naphtha, or white spirits” (OR = 1.81) or “paint thinners or removers” (OR = 2.73). The terms “paint thinners and removers” refer interchangeably with mineral spirits, white spirit, naphtha, VM & P naphtha, Stoddard solvent, and Varsol or petroleum distillation fractions that typically boil between 95°C and 210°C. Paint thinners, mineral spirits, gasoline, and ‘other solvents’ were solvents used for cleaning paint brushes with considerable overlap of exposures among these solvents. To assess the possibility that multiple solvent exposures might explain the associations with UCTD, a series of hierarchical models was run with different combinations of these four solvents. The best models all contained paint thinners and removers and a significant adjusted OR greater than twofold. After adjustment for paint thinners, mineral spirits showed no association with UCTD. There was a nonsignificant estimated 3% increase in risk of UCTD with each additional year of confirmed exposure to mineral spirits. The authors concluded exposure to paint thinners or removers was the exposure accounting for the significant associations with UCTD. Because of the multiple mixtures, there is an inability to precisely classify exposure to specific solvents. Thus, it was not possible to determine the specific role of mineral spirits in this study.

The study reported by Garabrant et al. (2003) is similar in design and source to study population to that reported by Lacey et al. (1999), except the association of interest is systemic sclerosis (SSc or scleroderma) and solvent exposure among 660 female cases and 2227 controls in Michigan and Ohio. Paint thinners and removers were significantly associated with systemic sclerosis both by self-report of exposure (OR = 1.9) and after expert review of exposure (OR = 2.0) after adjustment for age and year of birth. The risk remained after adjusting for correlated exposures, but there was no association with duration of exposure. Significant associations were reported for self-reported exposure to mineral spirits, white spirits, or naphtha (OR = 1.5), but were not significant when exposure was assessed by expert review (OR = 1.2). Once paint thinners and removers were taken into account no other identifiable solvent was clearly associated with systemic sclerosis.

Clinical markers of disease were evaluated in two studies of painters. Yaqoob et al. (1993) compared 112 automotive paint sprayers (exposed to paint-based mixtures containing primarily toluene, xylene, and n-butyl alcohol) to 101 transmission shop workers exposed to mineral oils and 92 pressmen (internal controls) with minimal exposure to lubricating oils. Cumulative hydrocarbon exposures at work and home were estimated. Painters in group 1 had a significantly increased prevalence of renal impairment measured as elevated serum creatinine, microproteinuria, N-acetylglucosaminidase, λ-glutamyltransferase and leucine aminopeptidase compared to the normal ranges derived from group 3. The authors suggested that paint exposure may be associated with renal impairment and microproteinuria, but they also noted that it “was not possible to relate these exposures to any specific hydrocarbon.” Because of the homogeneity of tenure and exposure within each group, it was not possible to assess exposure-response relationships. Exposure to mineral spirits was not mentioned. Thus this study is indeterminate with regard to the association between mineral spirit exposure and renal impairment.

Beving et al. (1991) evaluated the association of clinical changes in hematological parameters to effects of paint and organic solvent exposure of 17 car painters and 28 car mechanics to 46 office workers. Organic solvents used were toluene, xylene, white spirit, butanol, methyl isobutyl ketone, isopropanol, butyl acetate, and ethyl acetate and exposures were much higher for painters than for mechanics. The authors considered it reasonable to assume workers in this study were not exposed recently to benzene, although benzene impurities were present in the past. Red blood cell (RBC) counts were significantly decreased for both exposed groups (more so for mechanics) and the mode of the size distribution of RBCs was significantly increased toward larger volumes in both exposed groups. RBC volume was significantly increased among painters but not mechanics. This study is indeterminate with regard to mineral spirits because exposure was to a mixture with unknown proportions of solvents. There was no exposure-response analysis, but the differences in the RBC parameters between painters and mechanics were small and did not appear related to the presumed large differences in solvent exposure.

No causal conclusions can be drawn from these clinical-marker studies because of the lack of supporting mechanistic data and because of serious limitations within them. Both studies made only simple statistical comparisons between groups, without adjusting for confounding factors. In both cases, the authors suggested that the groups were “similar” regarding potentially important background characteristics such as smoking, drinking, diet, and exercise, without providing data. However, this assertion seems doubtful in the study by Beving et al. (1991), because the nonexposed control group consisted of office workers that presumably had different socioeconomic characteristics. Similarly, Yaqoob et al. (1993) suggested that exposed and referent ent groups were similar in age and other factors, while providing limited data suggesting that painters were significantly younger than pressmen.

This lack of control for confounding could have seriously biased these relatively weak results, as demonstrated in the McNamee et al. (1994) study. Those results initially suggested a monotonic dose-response trend and a nonsignificant 70% increased risk among those most exposed to paint solvents. However, these associations were completely removed following adjustment for confounding.

There may have also been some residual confounding in the UCTD study (Lacey et al. 1999), and SSc study (Garabrant et al. 2003) because the authors controlled only for age, sex, and birth cohort, without sufficiently establishing the comparability of cases and controls. These studies have similar limitations as they are by the same authors and are part of the same case-control study of rheumatic diseases (Garabrant et al. 2003). The biggest limitation of these studies is the potential for selection bias (i.e., diagnostic bias) inherent in the design. Undifferentiated connective tissue disease is not an established condition. Rather, it is a catchall term for conditions that do not meet the specific criteria for other rheumatic diseases, but share some signs or symptoms of established CTD. There is a considerable degree of subjective assessment involved in assigning this diagnosis, which could easily be influenced by the numerous reports published in dermatology and rheumatology journals linking solvents and CTD (Lacey et al. 1999), thus resulting in an artificial over-representation of exposure among the cases, which could not be removed during the analysis. Because milder cases of SSc are less likely to seek medical care than more serious cases, selection bias and non-representative sampling was possible.

The above problem is exacerbated by the poor participation rate among this case group. The authors suggest that they may have captured 75% to 80% (or more) of all eligible UCTD and SSc cases in Ohio and Michigan. However, when hospital refusal, patient refusal, and inability to contact are combined, the participation rate is probably closer to 35% to 40%. This low participation rate increases the possibility of selecting a biased case group of “volunteers.” These volunteers are generally unrepresentative of all cases or the general population from which the controls were selected. Such individuals are often more concerned about environmental agents and may have a greater stimulus to recall past occupational exposures (i.e., recall bias). The expert review of exposures failed to confirm 46% of SSc cases and 38% of control exposures for the category of “other solvents.” Greater overreporting of self-reported exposures by cases is also confirmed for exposures to mineral spirits, white spirits, naphtha, gasoline, and other specific solvents. Garabrant et al. (2003) also comment on more “spontaneous” recall of exposures by cases than controls, which was partly confirmed by expert review. Interpretation of exposure is also difficult because exposures are not mutually exclusive, and “in many instances reflect overlapping sets of chemical constituents.” The higher interview rate for cases (97% to 99%) compared to controls (73% to 80%) also suggests the possibility of some form of volunteer bias. These types of information bias are likely to produce overestimates of risk.

In conclusion, relatively few studies have evaluated specific exposures to mineral spirits or paints and its effect on target organ toxicity. The available studies are essentially indeterminate and provide no clear evidence of impacts of mineral spirits on any organ or body system.

Neurotoxicity Studies

In recent years, there has been a growing concern regarding the possibility of profound and/or irreversible nervous system effects resulting from chronic, low-level exposure to hydrocarbon solvents including mineral spirits. It has been well recognized that acute exposure to high concentrations of hydrocarbon solvents can produce a variety of transient central nervous system (CNS) effects, including dizziness, lightheadedness, lack of coordination, headache, irritability, and narcosis. In fact, due to their short-term CNS effects (i.e., narcosis), many solvents were used as general anesthetics (Axelson et al. 1980).

The ability of hydrocarbon solvents to produce chronic, long-lasting, or irreversible changes in the CNS following long-term, low-level exposure is a much more complex question. It has been proposed that protection against the acute effects of hydrocarbon solvents will also protect against potential chronic effects and this approach seems consistent with experience (Dick, 1988; Grandjean, 1985). However, some investigators conclude that occupational exposure (possibly even at current allowable occupational exposure levels) to a broad range of solvents including white spirits can produce irreversible brain damage. This neuropsychiatric condition is commonly referred to as “Organo Psycho Syndrome (OPS)” or as “Chronic Toxic Encephalopathy (CTE)”. OPS is characterized by vague and subjective symptoms, including fatigue, depression, poor memory and concentration, lack of energy, and changes in personality.

In order to establish a causal relationship between exposure to an agent and an effect, specific causal guidelines should be met, such as specificity of association and a dose-response relationship. Most of the epidemiological studies of OPS provide little, if any, information of the level of solvent exposure (i.e., dose). In fact, it is seldom possible to determine the specific solvents to which the workers were exposed. Many of the findings described in Scandinavian cases of OPS are nonspecific and may be influenced by other factors such as age, stress, genetics, nutritional status, environmental toxicants, head trauma, psychiatric disorders, alcohol, and other medications. In fact, more recent epidemiological studies cast doubt on whether long-term, low-level occupational exposures to solvents cause any significant neurological dysfunction (Gamble 2000; Ridgway et al. 2003). Data from existing animal studies provide little additional insight. Most experimental animal studies with solvents including white spirits were conducted to determine their acute neurotoxic effects on the central nervous system rather than the potential chronic neurotoxic effects resulting from long-term exposure. The repeated-dose studies that have been conducted show no pathological or functional differences when compared to controls.

Clinical Experiments

A number of controlled human exposure studies have been conducted to evaluate the neurobehavioral effects of acute exposure to white spirits (Gamberale et al. 1975; Cohr et al. 1980). Gamberale et al. (1975) exposed male volunteers to 100, 200, 300, and 400 ppm of white spirit (17% aromatics) for four consecutive periods of 30 min and examined performance in neurobehavioral tests (perceptual speed, single reaction time, short-term memory, numerical ability, manual dexterity) starting after 10 min in each exposure period. No effect on the examined functions was observed in these experiments. However, in a separate experiment using the same tests, exposure to 680 ppm for 50 min resulted in a prolonged reaction time and effects on short-term memory. At this concentration of white spirits in inspiratory air, an alveolar air concentration was obtained which corresponded to values obtained in exposure to 400 ppm during light exercise. From these data, the authors concluded that occupational exposure to 400 ppm white spirit for short periods might affect the psychomotor and intelligence function of workers.

Cohr et al. (1980) exposed nine house painters (occupational exposure duration >10 years; mean age: 49 years) to 0, 50, and 100 ppm white spirit (17% aromatics) and 15 students (mean age: 23 years) to 0, 34, 100, 200, and 400 ppm white spirit for 7 h. Clinical neurological observations were made and symptoms recorded. Psychological tests were performed before and during exposure. Increases in CNS symptoms (headache, tired-ness, and giddiness) were reported by the painters during exposure to 100 ppm. In painters, the only neuropsychological parameter that changed significantly during exposure to 50 ppm was reduction in short-term memory. The performance of the painters in the long-term memory function test was too poor to allow an evaluation. In the student group, CNS effects were only observed at concentrations of 100 ppm and higher where reaction time and attention function were decreased. Manual dexterity was affected by exposure to 200 and 400 ppm. Reduced long-term memory only occurred at 400 ppm. It is not clear whether the differences in performance between these two groups was due to differences in age, an age-dependent increase in sensitivity, or an occupational exposure–dependent increase in sensitivity.

There are two experimental studies of humans exposed for a short time to white spirits. Jarnberg et al. (1997) exposed nine healthy male volunteers to 298 mg/m³ white spirits in an exposure chamber for 2 h at a constant work load of 50 W on a bicycle. Before, during, and after exposure they rated severity of irritation and CNS-related symptoms: irritation of eyes, nose, throat or airways, headache, fatigue, nausea, dizziness, intoxication, difficulty in breathing. No significant irritation or CNS effects were detected, and the mean value of the rating ranged from ‘no effect’ to ’hardly any effect or discomfort.’

Pederson and Cohr (1984) exposed 12 volunteers for 6 h to white spirits at concentrations of about 0, 300, 600, and 1200 mg/m³ (0, 50, 100, and 200 ppm respectively). Results from self-administered questionnaires indicated no effect of exposure on symptoms of headache, dizziness, feeling of inebriation, visual disturbances, tremor, feeling of muscle weakness, impairment of coordination, paraesthesia of the skin, dryness of mucous membranes, loss of appetite, nausea, vomiting, diarrhea, or fatigue at the highest concentration of 200 ppm. Bebarta and DeWitt (2004) noted that in the Pederson and Cohr (1984) study there was no effect at 100 ppm, but exposures greater than 700 ppm can prolong reaction times. The source of the data on reaction is not clear because Pedersen and Cohr did not measure reaction time. The highest no effect level was 200 ppm, but this level is equal to the threshold limit value (TLV) of 100 ppm during light-moderate work. Bebarta and DeWitt (2004) also commented that in the study by Hane et al. (1977) house painters exposed to Stoddard solvent had slowed reaction times and attention deficits similar to those among workers exposed to solvents such as carbon disulfide and TCE. These exposure effects were observed, but exposure was to several hundred ppm of hydrocarbon solvents that were not identified. If Stoddard solvent causes symptoms, the no adverse effect level appears to be above the current TLV.

Epidemiology Studies

One epidemiological study which evaluated workers exposed solely to white spirits was conducted by Bazylewicz-Walczak et al. (1990). This study evaluated 226 workers (202 females, 24 males) exposed to white spirits (composition not specified) while gluing footwear elements in a cross-sectional study with 102 controls (91 females, 11 males) drawn from hosiery plants. Based on company records, white spirit concentrations were estimated to have been close to 85 ppm for the past 13 years; however, they may have been much higher in previous years. The participants were given a neuropsychological test battery (seven tests for intellectual functions and five tests for psychomotor performance). The performance of the exposed groups (as a total) was significantly worse with regard to four of the tests for intellectual functioning and three of the tests for psychomotor performance as compared to controls. The affected variables were correctness of perception and reproduction of visual material, projection of spatial relationships, concentration, speed of reactions to single and complex light stimuli, and manual dexterity. Variables such as simple and complex reaction time and coordination were found to deteriorate with duration of exposure. However, the conclusions of this study should be considered carefully in view of the deficiencies in study design. The exposure data in this study are poorly reported. No information on frequency and location of sampling was provided. It is also quite possible that high-peak exposures may have occurred on several occasions during the exposure period. In addition, it is not clear whether confounding factors such as alcohol consumption and premorbid intelligence were adequately taken into account.

A more recent cross-sectional study was conducted by Spurgeon et al. (1994). In this study, 110 paint makers exposed to solvents were compared with 110 age-matched controls for outcome measures designed to assess cognitive performance and mental health. Hygiene data available for the paint makers allowed the development of individual indices of solvent exposure and the analyses of health effects in relation to both duration and intensity of exposure. However, the indices of solvent exposure were based on exposures to solvent mixtures, not to individual solvents. The workers in this study were exposed to mixtures containing predominantly white spirit, toluene, xylene, methyl ethyl ketone, and methyl isobutyl ketone. Other substances present were acetone, isobutyl alcohol, n-butyl acetone, and n-butyl alcohol. No effects on cognitive functioning or mental health were found in the paint makers. For most of their period of employment, the exposures of the paint makers were below current occupational exposure limits. The results were interpreted as providing support that long-term exposure at or below current compliance levels does not result in damage to the CNS. Limitations of the exposure data in this study must be considered.

In conclusion, acute CNS effects following high-level exposure to white spirits (e.g., dizziness, headaches, etc.) have long been recognized. However, the ability of hydrocarbon solvents to produce chronic, long-lasting, or irreversible changes in the CNS following long-term, low-level exposure is a much more complex question. It is evident that resolution of this controversial issue will come only through the conduct of well-designed and controlled clinical epidemiologic studies, especially considering the absence of an appropriate animal model (Bruckner and Warren 2001). Reviews conducted by experts in the field such as Rosenberg (1995) and Schaumburg and Spencer (2000) have concluded that the current literature does not support chronic, low-level solvent exposure as a cause of symptomatic CNS or peripheral nervous system (PNS) dysfunction. This does not preclude, however, the possibility that such exposure can be associated with subclinical cognitive dysfunction in the form of slight psychomotor and attentional deficits. Gamble (2000) also conducted a weight of evidence review of studies investigating the effect of chronic exposure to hydrocarbon solvents in occupational cohorts. Results indicated no consistent pattern of decreased neurobehavioral function either in comparing exposed workers and unexposed controls, or in comparing high- versus low-exposure groups. Even when statistically significant associations occurred, the effects were often subtle and not adverse. The preponderance of the evidence suggested that long-term human exposure to hydrocarbon solvents at or below the occupational exposure recommendations does not result in adverse neurobehavioral effects. ECETOC (1996) similarly concluded, “there is no basis for a neurological syndrome in man that is causally related to low level organic solvent exposure (as defined by recent or current OELs).”

Reproductive/Developmental Toxicity Studies

Very few studies evaluated reproductive or developmental endpoints specifically for painters or those exposed to mineral spirits. One case-control study reported twice as much exposure to paint thinners and paint strippers among almost 2000 hospital-confirmed cases of spontaneous abortion, compared to randomly selected control births from the same county (Windham et al. 1991). The OR for exposure to oil-based paints was 0.96. Both oil paints and thinners were assumed to contain aliphatic solvents and paint removers to contain methylene chloride. There was some overlap between use of paint thinners and removers as well as with halogenated solvents perchloethylene and trichloroethylene. There was an inverse exposure-response trend for aliphatic solvents, with ORs of 2.3 and 1.4 for exposures less and more than 10 h/week, respectively. The inability to define adequately exposure to mineral spirits and the potential confounding effect of other exposures indicate these results are indeterminate for causal association.

Another case-control study compared the occupations (as listed on the birth certificate) of parents of children with CNS or digestive malformations to the parental occupations of children with other malformations. The more than 700 cases and 1700 controls were all identified from a Danish birth defects registry. The objective was to test the hypothesis that organic solvent exposure during pregnancy increased risk of congenital anomalies. These investigators reported an almost five times greater prevalence of employment as a painter among the fathers of children with CNS malformations. However, the quantification of the prevalence is unstable because it is based on only three exposed cases and there was no effect for maternal exposures (a more plausible association). The painters group was presumed to be the category in this study most exposed to solvents (Olsen 1983). These results are indeterminate based on the authors conclusion that exposure is poorly characterized, statistical control of potential confounding was very limited, and the finding of an association with fathers but not mothers was unexpected and an a posteriori finding that must await further research to determine whether it is an etiological association.

The other studies used cohort or pseudo-cohort designs to evaluate the birth outcomes of groups exposed to painting solvents. Heidam (1984) reported a 2 to 3 times higher risk of self-reported spontaneous abortion among women painters. These results are consistent with the study hypothesis, but the authors note that there is probably information bias as these elevated risks were not confirmed by either the hospital data or the second set of self-reported data. Further, they suggest confirmation is needed to support the hypothesis. The etiological nature of these results is indeterminate for organic solvents and for mineral spirits specifically, since no specific organic solvent was identifiable in this study.

Daniell and Vaughan (1988) found a nonsignificant 40% increased risk of low birth weight among children whose fathers were painters compared to general population controls. The calculated risk level was reduced to a risk ratio (RR) of 1.1 when compared to electrician controls that have a better match on socioeconomic status. There was no evidence for increased risk of prematurity, low Apgar scores, abnormal distribution of infant sex, or spontaneous fetal loss in most recent prior pregnancies. Exposure to mineral spirits is not known. The authors provide several reasons for the need of confirming these results before further interpretation, including misclassification of exposure based only on birth records, heterogeneity of exposure with only a small number of parents with job titles suggesting routine solvent exposure, lack of a mechanism to explain the finding, and possibly “chance results associated with multiple testing.” There was no control for smoking differences in either parent.

Similarly, Hoglund et al. (1992) reported slightly lower birth weight and length for the first-born children of 80 male spray painters exposed to several solvents below Swedish TLVs compared to 80 male electronics workers. Exposure was based on questionnaires answered by the man with his wife. This study is strengthened in that data were available on smoking and drinking. Despite about 20% missing smoking/drinking data on the women, stratified analysis did not support the hypothesis that these confounders caused the reductions in weight, length, or duration of pregnancy among painters. There were no differences in serious complications of pregnancy, malformations, or clinical course after birth. The authors indicated that these results must be interpreted with caution because the number of children and the differences were small and biological mechanisms to explain how the fathers’ exposure could affect the mother and child are unknown. Exposure is based on job title as in the other studies, and lacks information relating to the specific solvent exposures.

The few limited studies do not permit any causal conclusions. All five studies evaluated multiple exposures and/or birth outcomes, greatly increasing the probability of chance findings. In fact, it is not certain that paternal exposures were of primary a priori interest in the three studies that reported effects among the children of male painters. None of the authors discussed any reasonable mechanisms by which paternal exposure would cause these outcomes, suggesting that these may have been a posteriori findings (Olsen 1983; Daniell and Vaughan 1988; Hoglund et al. 1992). Also, Olsen (1983) points out the lack of supporting evidence for an association between CNS malformation and paternal painting, stating that this finding may be “of a purely hypothetical character.”

Recall bias was also a problem in two of the studies (Heidam 1984; Windham et al. 1991), because exposure or disease was derived from subjective self-assessment. This recall bias may explain the Heidam (1984) findings of increased risk only when women recalled past miscarriages, not when hospital-confirmed cases were used. All five studies would also be subject to some degree of residual confounding, especially the study by Olsen (1983) that relied solely on the limited covariate data recorded in the registry.

All those exposed to paint solvents in these studies were also subject to other occupational exposures. Only Windham et al. (1991) elicited information specifically for exposure to “paint thinners,” a category that would include mineral spirits. However, at least some of the increased risk for this category could have been the result of uncontrolled confounding, because adjustments for confounding were made only for the larger category of “all aliphatic solvents,” not paint thinners. There is also strong evidence that this increased risk may have been due to biased recall.

Spontaneous abortion is a traumatic event and should stimulate more intense recall of exposure for case mothers than for randomly selected control mothers. There were several indications that this in fact occurred in the study by Windham et al. (1991). Although paint thinners were moderately associated with increased risk, no increased risk was associated with exposure to oil-based paints (OR = 0.96), an exposure that should be highly correlated. No increased risk was seen for exposure at home (adjusted OR = 0.82), possibly because home exposures were seen as less threatening than occupational ones. Finally, all of the increased risk from paint thinner exposure was among those who reported subjective symptoms, skin contact, or odors (the primary complaint), which are factors that would elicit more intense recall.

The authors suggest that the latter two inconsistencies are indicative of a dose-response relationship, because the magnitude of exposure would be higher at work and with skin contact, odors, or symptoms (Windham et al. 1991). In that case, one would expect household or non-symptomatic exposures to still increase risk, but to a lesser degree. Instead, the risks among those with these less intense exposures were routinely decreased (OR = 0.7–0.8), which would suggest bias, not a dose-response relationship.

In conclusion, relatively few studies have evaluated specific exposures to white spirits or painting and reproductive/ developmental toxicity. The available studies provide no clear evidence for any adverse birth outcomes after exposure to these agents.

Cancer Epidemiology Studies

Eleven studies investigated associations with cancer. Half of these specifically evaluated white spirits or mentioned this as the predominant exposure, while the other half investigated occupational exposure as a painter. Two studies followed painter cohorts and the rest used case-control (CC) or nested CC designs. A summary of the epidemiological cancer studies of those exposed to “white spirits” or painting is presented in Table 10.

Hemminki et al. (1981) studied maternal and paternal occupations of Finnish children less than 15-years of age with cancer between 1959 and 1975. Cancers were classified into three categories: brain tumors, leukemia, and all others. Analysis of 145 paternal occupations showed six significantly increased ORs among 5 occupations, 1 of which was painting. There were 14 brain tumors for a nonsignificant (p < .10) OR of 2.59 over the entire study period. Subgroup analysis for the period 1969 to 1975 showed seven brain tumor cases and a significant OR (p < .05) of 5.00. For all malignancies the overall OR was 1.39 but was 2.75 and significant in 1969 to 1975. There were no significant excess risks for leukemia and other tumors either overall or for the later time period. Detailed maternal occupations at pregnancy included 98 occupations with significant ORs among 4 occupations. The authors concluded some of the occupations with significant ORs involved possible exposure to harmful chemicals (chromium and various solvents for painters) “although chance correlations cannot be excluded.” Given that there were over 5000 comparisons, more than 250 significant findings would be expected by chance alone in this part of the analysis. The authors concluded that even if the findings “could be biologically plausible, there is no way to exclude chance correlations” and other studies are needed.

Hardell et al. (1984) studied possible etiological agents for primary liver cancer in a case-control study of 98 cases and 200 controls in Sweden from 1974 to 1981. Exposure information was obtained from close relatives and jobs with exposure to organic solvents included repairers of cars and machinery, painters, and cleaners. The specific solvents included thinners, turpentine, and white spirit. The OR was 1.8 (0.99 to 3.4) for cases with high-grade exposure (continuous for more than 1 week or repeated brief exposures totaling 1 month or more) and “no major differences . . . between cases and controls” regarding low-grade exposure. But because of the “low number of exposed individuals, the differences could be explained by random variation.” Potential confounding effects of hepatitis B virus and cirrhosis could not be evaluated. Alcohol consumption was an important risk factor and was consumed more by cases than controls. Thus alcohol is a confounder, as it does not appear the risk ratios were adjusted for differences in alcohol consumption. There is an apparent discrepancy in results for organic solvents between Tables II and III in the publication. Calculated results from these two tables, showing the discrepancies, are shown below:

Low grade exposure over a lifetime is quite low (<1 month) and numbers are quite small, but these calculated risk ratios suggest there is no difference in risk between those with high and low-grade exposures. Further, it is not possible to estimate the attributable risk associated with mineral spirits relative to other organic solvents in this study. We suggest these results are indeterminate with regard to liver cancer and exposure to mineral spirits and need confirmation from other studies.

Wilcosky et al. (1984) conducted a nested case-control of several cancers showing excess mortality (stomach, prostate, lymphosarcoma, and lymphatic leukemia and respiratory cancer because of possible association with benzene) from 1964 to 1973 in a cohort of rubber workers. Jobs were classified on the basis of potential exposure to solvents, and then potential exposures to 20 different solvents used in the plant were related to each job over time. The only significant associations were for CCl₄ and CS₂ and lymphosarcoma and lymphatic leukemia. Occupational exposure greater than 1 year to VM&P naphtha and specialty napthas was negatively associated with respiratory cancer, and there were no significant associations with the other cancers. The authors suggest cautious interpretation even for the positive significant findings because of the large number of comparisons, the modest number of cancer cases, and possible biases (confounding due to high correlation between different solvents, probable negative biases due to exposure misclassification and lack of smoking data for respiratory cancer). Alternatively, the negative associations were said to possibly be due to chance. Thus this study is indeterminate at best.

Matanoski et al. (1986) conducted a cohort study of union painters and allied tradesman in four states in the United States from 1975 to 1979. The study population was stratified so results from the group of largely painters was presented separately, using the entire cohort as the reference population. Standardized mortality ratio (SMRs) for all 927 malignancies was elevated 10% (1.03 to 1.17), with excess coming primarily from 326 cases of lung cancer with an SMR of 1.18 (1.06 to 1.32) and 50 cases of stomach cancer with an SMR of 1.36 (1.01 to 1.80). No other causes of death were significantly elevated, although liver, bladder, and kidney had SMRs of 1.56, 1.26, and 1.41 with 20, 40, and 27 cases, respectively. There was a deficit in brain cancers with an SMR of 0.55 (9 cases). The use of the total population as the standard has the advantage of reducing (but not eliminating as suggested by the authors) possible biases due to differences between the cohort and U.S. male population in race, demography, and personal characteristics. A major limitation of this study is the lack of information regarding exposure, both as to duration and solvents. As the authors note, the components in paint are almost limitless and changing, and the mineral spirit composition is not known. Asbestos was in spackle in the past, and potentially could confound the association with lung cancer. The role of mineral spirits in these associations is indeterminate without the additional studies called for by the authors.

Lindquist et al. (1987) conducted a hospital-based case-control study of acute leukemia in Sweden. Controls were matched on age and sex from the population register. Both occupational and nonprofessional exposure to paints and solvents were determined by questionnaire. Skin exposure was defined as skin cleaning with aromatic and aliphatic solvents. Out of 125 cases and controls, there were 13 cases and 1 control who had been painters for an OR of 13 (2 to 554). The solvents used by painters were mixtures of aromatic and aliphatic hydrocarbons that the authors considered contained benzene as a contaminant. Organic solvents are used in many different professions, and analysis of these exposures indicated 26 cases and 7 controls were occupationally exposed to paints and/or solvents and/or glues for an OR of 3.7 (1.6 to 10). This is a heterogeneous group of occupations, half being painters and the rest only 1 occupation except 3 car repairmen, 4 woodworkers, and 2 train/bus wagon repairman and shoe factory worker each. There were 18 cases and 6 controls using organic solvents for daily skin cleaning, for an OR of 3.0 (1.14 to 9.24). The solvents used were white spirit (17% to 22% aromatic hydrocarbons) and/or gasoline with a high aliphatic hydrocarbon content and about 5% aromatic hydrocarbons composed of toluene, xylenes, trimethylbenzenes, and benzene. There was no association for nonprofessional painting. The authors concluded there was a greatly increased risk of developing acute leukemia after exposure to organic solvents. This study did report duration of exposure, but exposure was based on job and not on exposure to specific solvents. Benzene, toluene, and xylene were mentioned in the authors’ discussion. The role of mineral spirits appears indeterminate without further investigation in other studies to determine individual exposure to mineral spirits. Benzene is a likely confounder.

The largest of these studies was conducted by Siemiatycki et al. (1987a). In this population-based case-control study, a total of 739 men were classified as having been potentially exposed to “mineral spirits.” The term “mineral spirits” was used broadly to include white spirits, Stoddard solvent, VM&P naphtha, rubber solvent, benzene, and ligroin (80% to 90% aliphatics, 1% to 20% aromatics). The main occupations were construction painters (21%), mechanics and repairmen (20%), and metal machining (5%). For those with more than 30 cases, exposure was dichotomized by duration with a 10-year cut point and median based on concentration, frequency, and probability. Long-high exposure to mineral spirits produced adjusted ORs of 1.7 (1.2 to 2.3) for squamous cell lung cancer and 1.8 (1.3 to 2.6) for prostate cancer; short-high exposure produced a 1.7 (1.1 to 2.8) OR for prostate cancer. When there were fewer than 20 exposed cases, there were two exposure subgroups (substantial and nonsubstantial) based on cumulative exposure, which combines level and duration. Substantial exposure produced an OR of 2.0 (1.0 to 4.1) for Hodgkin’s lymphoma. The authors acknowledged the possibility of significant finding for both positive and negative associations because of the large number of comparisons and because the purpose was to generate hypotheses. The in-depth analysis by exposure level was an attempt to help separate false positives from true positives. They considered the etiologic role of mineral spirits for lung cancer to be “realistic,” but biological plausibility for prostate and lymphoma etiologic was not “self-evident.” Although increased relative risks for certain cancers were observed in this and other studies, the International Agency for Research in Cancer (IARC) concluded that the results of these studies could not be evaluated with regard to the petroleum solvents themselves. The overall IARC was that petroleum solvents are not classifiable as to their carcinogenicity to humans (IARC Group 3).

Persson et al. (1989) conducted a hospital-based case-control study of Hodgkin’s disease (HD) and non-Hodgkin lymphoma (NHL) and occupational exposures. Controls were drawn from the Swedish population registry from the catchment area of the hospital. Exposure was determined by questionnaire. A second study of similar design was then conducted in a different region in Sweden to confirm or refute the earlier findings (Persson et al. 1993). Analysis consisted of calculation of crude ORs followed by logistic regressions for exposure with twofold or greater risk and adjusting for “identified risk factors,” which consisted of age at diagnosis, exposures of a priori interest, and exposures with preventive effect in the crude analyses. The authors concluded both studies suggested solvent exposures were risk factors for malignant lymphomas. Analysis of crude ORs suggested white spirit was a risk factor also. The comparative results of the two studies are summarized:

Risks associated with solvents and white spirit specifically are considerably less and non-significant in the second study. The confidence intervals are quite wide indicating the relatively small sample size and instability of the risk estimates. There is little difference between the crude and adjusted OR, suggesting little confounding or inability of logistic regression to account for limitations in study design suggested below.

Several factors limit interpretations of the Persson et al. (1989, 1993) results.

• Although the questionnaire appeared to be self-administered, the quantitative classification of exposure was not done blindly as the controls (and therefore classification) was from earlier studies where these referents were also used. This could bias exposure estimates upward.

• The controls were population based whereas the cases were hospital-based, which increases the potential for recall bias as cases may be more likely to remember than healthy controls. Information was not provided regarding characteristics of cases and controls so the similarity of cases and controls cannot be evaluated.

• The authors concluded there appears to be an association with solvents, but the meaning of this term is unclear and is non-specific for any particular solvent or group of solvents. For example, in the first study (Persson et al. 1989), there were 18 HD cases exposed to solvents, 14 exposed to solvent intensities 2 to 5, and 23 exposed to 4 individual solvents (thinner, white spirit, trichloroethylene, and styrene) at intensities 2 to 5. And there were 2 HD cases who were painters. These numbers don’t seem to add up, making it unclear how many were exposed and to what. Apparently painters were not considered to be exposed to solvents in this study.

• Adjusted ORs used 90% confidence intervals, so the customary p < .05 level of significance was replaced with p < .10 level of significance. Ninety-five percent confidence intervals were calculated for crude OR and white spirit exposure, and all were non-significant except for NHL in the first study. If the customary p < .05 level of significance had been chosen, adjusted OR would not have been statistically significant for either HD or NHL, and the associations with solvents could be due to chance.

• The etiologic role of solvents is somewhat ambiguous as there is little or no increased risk at higher exposure (comparing solvent exposure to high intensity solvent exposure). The exception is for NHL in the second study where high intensity OR is 1.7 relative to all solvent OR of 1.2.

Guberan et al. (1989) studied cancer incidence in a cohort of 1916 Geneva painters identified in the 1970 census. Significant excesses were observed for cancers of the gallbladder, lung, testis, and bladder. The authors concluded the highly significant excess of lung cancer was possibly related to confounding exposure to the carcinogens asbestos and zinc chromate. Further duration and exposure levels, as well as smoking levels, are unknown, so the role of painting is unknown. There were only three cases of gallbladder cancer. There were 13 cases of bladder cancer and although there is some support from other studies of painters showing an excess, the authors postulated the excess might be due to smoking or exposure to an unknown bladder carcinogen, or both. There were five cases of testicular cancer, and because this is a new finding a “chance excess should be considered.” These results are indeterminate with regard to mineral spirits as the only connection is via job at one point in time. There are no data on solvents composition, duration, or intensity of exposure. Further, confounding exposures (e.g., asbestos, smoking) are probable but speculative.

Bethwaite et al. (1990) conducted a series of case-control studies nested within a cohort of painters registered with the New Zealand Cancer Registry. This study is similar to a cohort study in that RRs for major causes of cancer are reported. The difference is that the controls are registrants for all other cancer sites rather than a national population as in the cohort design. These controls are advantageous in that information bias is minimized and selection bias reduced from incomplete registration. A possible disadvantage is selection bias if exposure is associated with more than one cancer. Occupation is the current or most recent job at the time of registration. The main findings were significantly increased ORs of 1.53 (1.10 to 2.14) for bladder cancer and 1.95 (1.05 to 3.65) for multiple myeloma. The authors noted the lack of excess risk for such sites as lung, gastrointestinal, liver, gallbladder, and leukemia. Organic solvent exposure was from use as cleaners, with white spirit being commonly used by house painters. Other potential confounding exposures included aromatic solvents (e.g., toluene, xylenes), range of adhesives and degreasing agents commonly containing trichloroethylene, n-hexane, methyl ethyl ketone, or benzene as an impurity. Major limitations of this study were similar to others, that “detailed knowledge on the specific risks posed by individual solvents is limited” and the lack of information on potential confounding factors such as smoking. The excess bladder cancer for example might be due in part to the higher smoking commonly found among painters.

Parent et al. (2000) conducted a case-control study of esophageal cancer and workplace exposures in Montreal. This design was similar to that used by Siemiatycki et al. (1987a) with some improvements in exposure assessment. There were 99 cases with a total of 1066 combined population and cancer controls. Logistic regressions adjusted for age, birthplace, education, respondent, smoking, alcohol, and ß-carotene intake. Exposure analysis was limited to 35 occupational agents, one of which was mineral spirits. Two types of mineral spirits exposure were coded; one (generally before 1970) included benzene, toluene, and xylene, whereas the second formulation did not include these constituents. Because preliminary analyses indicated no association with the later formulation, the presented results are for the mineral spirit formulation containing benzene, toluene, and xylene. The OR for any, nonsubstantial, and substantial exposures were 1.1 (0.6 to 2.0), 1.6 (0.7 to 3.7), and 0.9 (0.4 to 1.9). For only squamous cell cancer the OR were somewhat higher at 1.9 (1.0 to 3.7), 2.6 (1.0 to 6.6), and 1.6 (0.7 to 3.6), respectively. In both instances, there was an apparent lack of an exposure-response trend as unsubstantial exposure posed a higher risk than substantial exposure and neither were significantly different than no risk (OR = 1.0). The authors considered the association with the early formulation of mineral spirits as equivocal. This population-based study and others from the Montreal group are model studies for this design because of careful assessment of past exposures, adjustments for potential confounders, and selection of controls.

Discussion of Cancer Studies

Excess risk was most commonly reported for lung cancer, which was significantly elevated in the painter cohorts (Matanoski et al. 1986; Guberan et al. 1989) and in the large case-control study performed by Siemiatycki et al. (1987a). Overall lung cancer risks were generally weak, with risk ratios (RR) ranging from 1.2 to 1.5 (Table 10). No clear dose-response relationships were reported, although Siemiatycki et al. (1987a) noted the highest risk among those in the highest exposure category (OR = 1.7, 95% CI = 1.2–2.3). Wilcosky et al. (1984) and Bethwaite et al. (1990) also investigated lung cancer, but reported nonsignificant OR of 0.54 and 1.12, respectively.

One cohort and one case-control study found significantly elevated bladder cancer risks for painters (Guberan et al. 1989; Bethwaite et al. 1990). However, these associations were relatively weak (1.5 to 1.7) and were not replicated in the other studies that investigated urinary bladder cancer (Matanoski et al. 1986; Siemiatycki et al. 1987a).

Several lymphohematopoietic (LH) malignancies were also elevated, but these did not demonstrate a consistent pattern. Persson et al. (1989) reported a significant three times greater risk of exposure to white spirits among non-Hodgkin’s lymphoma (NHL) cases, but a later study by the same authors found only a non-significant 60% increase (Persson et al. 1993). Other studies that evaluated NHL failed to report any substantially increased risk (Wilcosky et al. 1984; Siemiatycki et al. 1987a; Guberan et al. 1989; Bethwaite et al. 1990). Hodgkin’s disease (HD) risk was elevated in two studies (Persson et al. 1989; Siemiatycki et al. 1987a), with one reporting a marginally significant two times greater risk among those with “substantial” exposure (Siemiatycki et al. 1987a). However, Persson et al. (1993) reported a much lower risk in their later study (OR = 1.4) and Bethwaite et al. (1990) found no increased HD risk (OR = 0.95), based on nine exposed cases. Two studies reported substantially increased leukemia risks among exposed workers (Wilcosky et al. 1984; Lindquist et al. 1987), but several others failed to replicate these findings (Matanoski et al. 1986; Guberan et al. 1989; Bethwaite et al. 1990).

Parent et al. (2000) conducted a detailed analysis of esophageal cancer in a follow-up of the original study by Siemiatycki et al. (1987b). There were 17 cases exposed to mineral spirits. Sixteen of these were squamous cell carcinomas (SCCs) and a twofold excess risk. There was an inverse dose-response relationship as OR was 2.6 (CI = 1.0–6.6) in the nonsubstantial exposure group and 1.6 (CI = 0.7–3.6) in the exposure group with substantial exposure to mineral spirits.

Most studies noted only weak effects from exposure, which could be the result of uncontrolled confounding or bias. Only Siemiatycki et al. (1987a) and Parent et al. (2000) adequately controlled for smoking, occupational risk factors (e.g., chromate or nickel exposure), or similar potential confounders. The lack of control for smoking would be especially problematic for associations with bladder or lung cancer, because this habit is a strong risk factor for both of these diseases and is likely to be more prevalent among blue-collar workers (e.g., painters) than among the general population. Guberan et al. (1989) noted no increased lung or bladder cancer RR for electricians, who are similar demographically to painters, but less exposed to solvents. However, the standardized mortality/morbidity ratios calculated in this study are not directly comparable across cohorts, especially considering that the electricians had a younger age pattern than the painters (Guberan et al. 1989).

Lindquist et al. (1987) reported a significant acute leukemia OR of 13 for painters. This association could be at least partly due to uncontrolled confounding, because Lindquist et al. (1987) controlled only for age and gender in their painter analyses. Past exposure to benzene may have caused some of this elevation, because most of the painters worked during time periods when higher-level benzene exposure was more common and because only acute myelogenous leukemia was significantly associated with solvent exposure in this study. Recall bias is also possible, because cancer cases were compared to healthy controls and exposure was determined subjectively by questionnaire. In any case, this association should probably be considered an atypical or chance finding, unless substantially more corroborative evidence is uncovered. This finding has not been replicated in other painter studies (Matanoski et al. 1986; Guberan et al. 1989; Bethwaite et al. 1990) or in cohort studies of paint manufacturing (Morgan et al. 1981; Lundberg and Milatou-Smith 1998).

The strongest methodological study in this body of literature, performed by Siemiatycki et al. (1987a), includes the separate analysis of esophageal cancer by Parent et al. (2000). These authors simultaneously controlled for multiple confounders and used cancer controls to limit recall bias. Also, individual exposures were assigned by chemists/hygienists based on occupation, not by the subjects themselves. However, this research is extremely exploratory, a fact that is acknowledged by the authors. Furthermore, only weak associations were noted for mineral spirits, and no clear dose-response trends were reported for this exposure. The statistically significant findings for mineral spirits may simply reflect the greater number of cases exposed to this common chemical class and the authors’ choice of a one-sided test of significance.

Another common problem in these studies is the inadequate and inconsistent classification of exposure and lack of exposure-response analysis. Often exposure to mineral spirits is part of overall organic solvent exposure, which, if characterized at all, is often reported as a mixture likely to include benzene and other aromatic solvents. Parent et al. (2000) provided an example of the potential confounding effect of organic solvent mixtures when they found no risk of esophageal cancer associated with a formulation of mineral spirits without benzene, toluene, and xylene, but suggestive increased risks when these aromatics were present. Being a painter was often a surrogate for organic solvent and mineral spirit exposure, but without documentation of specific solvent exposures. Although some studies of painters are included, exposure by job is less satisfactory than exposure to specific agents.

In conclusion, this literature contains some suggested associations between mineral spirits and several cancers. However, these associations are typically weak and are not consistent across studies. Individuals exposed to mineral spirits in these studies would also have been subject to other occupational exposures. Therefore, in the absence of substantially more epidemiological or mechanistic evidence, these data provide no strong indication that mineral spirits cause any cancer.

Summary of the Epidemiology Findings

Relatively few epidemiological studies have evaluated individuals exposed specifically to mineral spirits. There are considerably more studies that have evaluated general solvent exposure. However, these would be even less specific to mineral spirits than the studies evaluated in this review, making causal conclusions even more problematic.

Neurological impairment is the most commonly studied health endpoint and has been adequately summarized in a recent technical report (ECETOC 1996). The acute central nervous system effects of high-level exposure to mineral spirits (e.g., dizziness, headaches, etc.) have long been recognized. However, the ability of hydrocarbon solvents to produce chronic, long-lasting, or irreversible changes in the CNS following long-term, low-level exposure is a much more complex question. The evidence suggests that long-term human exposure to hydrocarbon solvents at or below current occupational exposure recommendations does not result in adverse neurobehavioral effects. Reproductive and/or developmental effects and organ impairment have been reported by a few investigators, but the sparseness, inconsistency, and overall low quality of these studies do not suggest any causal relationships. Several associations with cancer have also been reported, but these findings are generally weak, inconsistent as to cancer site, and may be due to confounding or bias. Therefore, the existing data do not support the thesis that mineral spirits cause non-neurological adverse impacts in people.

KEY CONCLUSIONS

Toxicological Studies

• Mineral spirits have a low order of acute toxicity by the oral, dermal, and inhalation routes of exposure.

• Aspiration of mineral spirits into the lungs may produce serious damage leading to bronchopneumonia that may be fatal within 24 h.

• In repeated dose studies, the most common findings are kidney effects in male rats, consistent with α _2u-globulin–mediated nephropathy.

• The data suggest that mineral spirits are not reproductive or developmental toxicants.

• The weight of evidence suggests that mineral spirits are not genotoxic.

• In a 2-year carcinogenicity study with mineral spirits (ASTM Type II Class C Stoddard Solvent), there was some evidence of carcinogenic activity in male rats based on increased incidences of neoplasms in the adrenal medulla. The mechanism by which Stoddard Solvent IIC produces adrenal neoplasms remains to be determined. There was no evidence of carcinogenic activity in female rats or in male mice. In female mice there was “equivocal” evidence based on an increased incidence of liver tumors.

Human Observations

• In humans, acute high-level inhalation exposure to mineral spirits can produce a variety of CNS effects such as dizziness, headaches, and lack of coordination.

• At current occupational exposure levels, there is no compelling evidence that mineral spirits produce irreversible CNS effects, although this remains controversial.

• A causal relationship between exposure to mineral spirit and reproductive and/or developmental effects has not been demonstrated.

• Several associations with cancer have been reported, but these are generally weak, inconsistent as to cancer site, and may be due to confounding or bias.