Abstract
Prior studies have suggested that men with prostate cancer and psychiatric disorders (+Psy) have worse outcomes compared with those without (−Psy), particularly due to delayed diagnosis or reduced access to definitive treatment. In the current study, the toxicity and outcomes of men who were primarily diagnosed through prostate-specific antigen screening and who underwent definitive treatment with external beam radiation was investigated. The charts of 469 men diagnosed with prostate cancer from 2003 to 2010 were reviewed. The presence of +Psy was based on a Diagnostic and Statistical Manual of Mental Disorders–Fourth edition diagnosis of posttraumatic stress disorder, depression, schizophrenia, bipolar disorder, and/or generalized anxiety disorder. Kaplan–Meier analysis was used to analyze biochemical control, distant control, prostate cancer–specific survival, and overall survival. One hundred patients (21.3%) were identified as +Psy. At a median follow-up of 73 months, there were no differences regarding 6-year biochemical control (79.8% −Psy vs. 80.4% +Psy, p = .50) or 6-year distant metastatic-free survival (96.4% −Psy vs. 98.0% +Psy, p = .36). There were also no differences regarding the 6-year prostate cancer–specific survival (98.4% −Psy vs. 99.0% +Psy, p = .45) or 6-year overall survival (80.2% −Psy vs. 82.2% +Psy, p = .35). Short- and long-term genitourinary and gastrointestinal toxicities were similar between the groups. On multivariate analyses with propensity score adjustment, +Psy was not a significant predictor for toxicity, biochemical recurrence, or survival. The presence of +Psy was not associated with higher toxicity or worse clinical outcomes, suggesting that effective removal of screening and treatment barriers may reduce the survival disparities of these patients.
Introduction
Overall mortality rates in psychiatric patients are much greater than in the general population (Alström, 1942; Berren, Hill, Merikle, Gonzalez, & Santiago, 1994; Cuijpers et al., 2014; Harris & Barraclough, 1998; Odegard, 1952). Despite similar incidence, cancer mortality has been reported in some studies to be increased in patients with psychiatric disorders compared with the general population (Chang et al., 2014; Guan et al., 2013; Lawrence, Holman, Jablensky, Threlfall, & Fuller, 2000; Pinquart & Duberstein, 2010; Satin, Linden, & Phillips, 2009; Tran et al., 2009). On a behavioral level, cancer survival is often dependent on early diagnosis, access to care, and compliance with the treatment regimen. The disparity between incidence and mortality may be due to decreased or delayed diagnosis, lack of access to screening, lack of access to proper care, or decreased compliance with treatment. On a biological level, chronic activation of the hypothalamopituitary-adrenal axis (Reiche, Nunes, & Morimoto, 2004) or increase in proinflammatory cytokines in patients with mental illness, (Seruga, Zhang, Bernstein, & Tannock, 2008) may further influence cancer progression and outcome.
With the advent of prostate-specific antigen (PSA) test screening, early diagnosis along with assessing treatment response and recurrence of prostate cancer have become more precise. While the utility of PSA screening has recently come under question, it has nevertheless been proven to reduce prostate cancer mortality (Schröder, 2012). Despite these improvements, prostate cancer is persistently the second highest cause of cancer mortality among men, behind only lung cancer (American Cancer Society, 2014).
In the current study, the outcomes of men with or without major psychiatric disorders who were diagnosed with prostate cancer in the PSA era and who were treated with external beam radiation at the New York Harbor Veterans Affairs Hospital was investigated. Since these patients were already screened, had agreed to treatment, and were in an equal access center, these men, despite their psychiatric illness, were less likely to be affected by lack of access to screening or access to appropriate care.
Method
After approval by the New York Harbor Department of Veterans Affairs Institutional Review Board, the charts of 469 consecutive men who were diagnosed with prostate cancer and were treated with external beam radiation to a dose of 7,560 cGy or higher from 2003 to 2010 were reviewed. The radiation techniques have been previously described (Surapaneni, 2014). In general, patients treated from 2003 to 2006 were treated via three dimensional conformal radiation therapy. Starting in late 2006, there was a gradual transition to intensity modulated radiation therapy (IMRT) so that by 2008, all patients were being treated via IMRT. The radiation fields generally included the whole pelvis for patients with high-risk disease and the prostate ± seminal vesicles for low- and intermediate-risk disease.
The medical records were reviewed to identify a history of a Diagnostic and Statistical Manual of Mental Disorders–Fourth edition psychiatric disorder that was present at the time of the prostate biopsy. Patients were identified as having no psychiatric disorder including posttraumatic stress disorder, depression, schizophrenia, bipolar disorder, and generalized anxiety disorder. The latter five diagnoses were grouped broadly as being positive for a history of psychiatric disorder. Toxicity was evaluated both during and after treatment by analyzing medical records across multiple clinics in the Veterans Hospitals system. Toxicities were graded using the National Cancer Institute Common Terminology Criteria for Adverse Events version 3.0. Grade 1 toxicity corresponded to minimal side effects, Grade 2 toxicity corresponded to side effects requiring medications, Grade 3 corresponded to side effects requiring minor procedures, and Grade 4 toxicity corresponded to medical admission due to life-threatening complications.
Biochemical failure after radiation therapy was defined using the definition of PSA nadir +2ng/mL and calculated from the date of the completion of radiation therapy. Patient characteristics were analyzed and compared between those who were +Psy and −Psy using Chi-square, Mann–Whitney, and Fisher’s exact test where appropriate. Differences in toxicity between +Psy and −Psy were also compared via Fisher’s exact test. Survival data were gathered from the Veterans Hospital medical records as well as from social security records. Biochemical disease-free survival was analyzed using the Kaplan–Meier method and compared using the log-rank test. Patients were divided into risk groups based on National Cancer Care Network (NCCN) criteria (www.nccn.org). Propensity scores were also derived from a logistic regression model for the presence of a psychiatric disorder. The covariates included age, year of diagnosis, race, NCCN risk group, and androgen deprivation usage. Propensity scores were then used as continuous covariates in the multivariate logistic regression as well as multivariate Cox regression survival analyses. Multivariate logistic regression was used to determine predictors for Grades 3 to 4 toxicity. Univariate and multivariate Cox regression modeling was performed to determine the impact of covariates on biochemical outcome and overall survival. Covariates measured included age (continuous variable), race (Caucasian, Black, or Other), NCCN risk group (low, intermediate, high), radiation technique (three dimensional conformal radiation or intensity modulated radiation), and the presence or absence of a psychiatric disorder (yes or no). Statistical analysis was performed using SPSS version 21.0 (IBM Inc., Armonk, New York) and statistical significance was achieved with a p value < .05.
Results
There were 369 patients (78.7%) who were identified as having no known psychiatric disorder (−Psy). The remaining 100 patients (21.3%) were identified as having posttraumatic stress disorder (n = 51), depression (n = 29), schizophrenia (n = 13), bipolar disorder (n = 5), and generalized anxiety disorder (n = 2) (+Psy).
The median follow-up was 73 months (range 11-132 months) and 414 (88.3%) men were followed for a minimum of 3 years after completing radiation therapy treatments. The median follow-up for +Psy was 64 months versus 74 months for −Psy (Mann–Whitney U test, p = .38). The median number of days from diagnosis until initiation of radiation therapy was 122 days for −Psy and 117 days for +Psy (Mann–Whitney U test, p = .62).
With regard to patient characteristics, +Psy were a median of 9 years younger than −Psy (p < .001). There were not any significant differences between the two groups with regard to the year treated, race, risk group, and radiation technique. Additional details are available in Table 1.
Patient Characteristics.
Note. NCCN = National Cancer Care Network; 3DCRT = three dimensional conformal radiation therapy; IMRT = intensity modulated radiation therapy.
Mann–Whitney U test.
With regard to toxicity, there were no differences between +Psy and −Psy in both short- and long-term tolerance of radiation treatments (Table 2). On multivariate logistic regression with propensity score adjustment for late grade ≥3 genitourinary or gastrointestinal toxicity, there were no significant associations with late toxicity (Table 3).
Acute and Chronic Treatment Toxicity.
Note. GU = genitourinary; GI = gastrointestinal.
Multivariate Logistic Regression Analysis for Late Toxicity.
Note. GU = genitourinary; GI = gastrointestinal; OR = odds ratio; CI = confidence interval; NCCN = National Cancer Care Network; 3DCRT = three dimensional conformal radiation therapy; IMRT = intensity modulated radiation therapy.
There were a total of 90 biochemical failures and the 6-year biochemical recurrence-free survival was 79.8% for −Psy and 80.4% for +Psy (p = .50; Figure 1). On multivariate analysis, intermediate and high-risk disease were associated with a significantly higher risk for biochemical recurrence. The presence of +Psy was not associated with any differences in biochemical outcomes on multivariate analysis (hazard ratio [HR] = 0.96, 95% confidence interval [CI] [0.55, 1.66], p = .88, Table 4).
Biochemical control in prostate cancer patients with or without psychiatric disorders.
Univariate and Multivariate Analysis for Biochemical Recurrence Free Survival.
Note. HR = hazard ratio; CI = confidence interval; NCCN = National Cancer Care Network; 3DCRT = three dimensional conformal radiation therapy; IMRT = intensity modulated radiation therapy.
There were a total of 17 patients who developed distant metastatic disease. For these patients, the median time to distant metastatic disease progression was 32.5 months from the end of radiation treatment. The 6-year distant metastatic disease-free survival was 96.4% for −Psy and 98.0% for +Psy (p = .36).
With regard to overall survival, the 6-year overall survival was 80.2% for −Psy and 82.2% for +Psy (p = .35). There were a total of 10 prostate cancer–related deaths, nine of which were in −Psy and one in +Psy. The 6-year prostate cancer–specific survival was 98.4% for −Psy versus 99.0% for +Psy (p = .46). On multivariate analysis for overall survival, only age >70 was significantly associated with mortality. However, +Psy was not associated with any difference in overall survival (HR = 1.05, 95% CI [0.63, 1.76], p = .86, Table 5).
Univariate and Multivariate Analysis for Overall Survival.
Note. HR = hazard ratio; CI = confidence interval; NCCN = National Cancer Care Network; 3DCRT = three dimensional conformal radiation therapy; IMRT = intensity modulated radiation therapy.
Discussion
The current unique retrospective study compared prostate cancer–specific endpoints after primary external beam radiotherapy in patients with or without psychiatric disorders. There were no differences noted in biochemical control, overall survival, prostate cancer–specific survival, distant metastasis-free survival, or toxicity between men with or without psychiatric disorders at a median follow-up of 73 months. On multivariate analysis, the presence of psychiatric disease was not a significant predictor of late toxicity or biochemical recurrence.
The association of increased mortality in patients with mental disorders dates back to the mid-19th century (Farr, 1841) and has been well established in multiple studies over the past 150 years (Alström, 1942; Berren et al., 1994; Cuijpers et al., 2014; Harris & Barraclough, 1998; Odegard, 1952). Several recent studies have corroborated increased cancer-related mortality in this subgroup of patients (Chang et al., 2014; Guan et al., 2013; Lawrence et al., 2000; Pinquart & Duberstein, 2010; Satin et al., 2009; Tran et al., 2009). Proposed reasons for this survival discrepancy include delayed diagnosis or misdiagnosis of cancers, decreased access to care, underutilization or noncompliance with treatment modalities due to inability to cope with treatment regimens, and failure to pursue more aggressive treatment options (Colleoni et al., 2000; Howard et al., 2010).
Specifically in prostate cancer, there is a recent Surveillance, Epidemiology, and End Results Program (SEER) Medicare study that reported increased mortality in depressed men with prostate cancer. In that report, they noted that depressed men were less likely to choose definitive therapy. Even among those who did choose definitive therapy, the odds ratio for overall survival was 0.68 to 0.77, depending on the risk grouping (p value ranging from <.001 to .006; Prasad et al., 2014). This is in apparent contrast to the current study in which there were no differences in any outcome endpoint, including not only survival but also biochemical control and distant metastatic disease. One possible explanation is that the SEER-Medicare study only analyzed overall survival and not prostate cancer–specific survival. As the authors noted, men in their study with depression were more likely be older, lower income, and have more comorbidities. It is possible that the overall mortality differences noted in their report reflect the impact of these comorbidities rather than any differences in prostate cancer–specific mortality. A large population-based study from Western Australia examining 135,442 patients with cancer reported that the overall mortality rates were higher in +Psy patients compared with the rest of the population, with overall mortality ratio of 1.52 (95% CI [1.45, 1.60]) in men and 1.29 (95% CI [1.22, 1.36]) in women, respectively (Kisely, Crowe, & Lawrence, 2013). It was also noted that the +Psy patients were more likely to present with metastatic disease and less likely to undergo definitive treatments. Again, one may attribute much of the differences in mortality to delayed diagnosis or lack of access to medical care.
The possibility of delayed diagnosis affecting survival in patients with cancer have been reported in multiple settings, including esophageal cancer in the U.S. Veterans system (O’rourke et al., 2008), multiple malignancies in one study from Bombay, India (Chakravorty, Chakravorty, Patel, DeSouza, & Doongaji, 1993) and breast cancer in a Connecticut tumor registry study (Desai, Bruce, & Kasl, 1999). In an attempt to account for the impact of delayed diagnosis, Chang et al. (2014) performed a registry-based study from patients in parts of London, England (Chang et al., 2014) in which they adjusted for stage at diagnosis and still reported that patients with +Psy had worse survival. It should be noted that Chang et al. (2014) did not have detailed data regarding receipt of treatment available. As a result, one cannot exclude the possibility that the worse survival outcomes for +Psy men identified by Chang et al. (2014) is actually a reflection of +Psy men having been less likely to undergo definitive treatment.
One of the benefits of the current study is that it has reduced the impact of delayed diagnosis by focusing on prostate cancer patients, who were generally diagnosed via PSA screening. PSA screening, though controversial due to the question of overdiagnosis, has been proven to result in earlier diagnoses of prostate cancer (Catalona, Smith, Ratliff, & Basler, 1993) and reduced prostate cancer mortality (Schröder, 2012). Due to PSA screening, the vast majority of the men included in the current study were diagnosed at an early stage, with ~70% being diagnosed with low-risk or intermediate-risk disease (Stages I-IIA). All men agreed to definitive treatment with radiotherapy, thereby reducing the potential impact of lack of access to care or definitive treatment. In the current setting, there were no differences in any outcome endpoint between those with or without psychiatric disorders. This lends support to the possibility of other confounding variables being the cause of the mortality differences noted in the aforementioned population-based studies. The challenges of using population-based studies in interpreting mortality endpoints has been reported in several studies to be a potentially flawed analysis due to unmeasured confounding variables (Giordano et al., 2008; Park, Lloyd, Decker, Wilson, & Yu, 2012).
In addition to the question of the impact of psychiatric disorders on mortality, no one has investigated whether these men exhibit any differences with regard to tolerance to treatment, particularly radiation therapy. A variety of clinical factors have been investigated as potential risk factors for increased toxicity for radiation. These include radiation dose (Peeters et al., 2006), radiation fields (Beard et al., 1997), medications such as anticoagulants (Vavassori et al., 2007), and the use of androgen deprivation therapy (Sanguineti et al., 2002). In the current study, the toxicity profiles were similar to prior reports (Kok, 2013; Zelefsky et al., 2002; Zelefsky et al., 2012) and there were no differences in toxicity between those with or without psychiatric disorders (Table 2). To our knowledge, this is the first study to investigate whether patients with psychiatric illnesses tolerate treatment any differently than those without any psychiatric disorders. Since the toxicity endpoints in the current study are physician reported, differences in toxicity may warrant further study using validated patient reported outcome measures.
There are several limitations to the current analysis. First, this is a retrospective review and the subject patient population may be biased as a highly selected group of patients who both agreed to undergo screening as well as treatment. Second, while the median follow-up is 73 months, this is likely insufficient follow-up to detect clinical differences in mortality rate in patients treated for localized prostate cancer. Third, the patient population in the current study consisted of >60% African American men, whose prostate cancers have been reported to be biologically different from Caucasians (Kim et al., 2011) and associated with increased mortality (Siegel, Naishadham, & Jemal, 2013). Although race was not a significant variable in the current analysis, confirmatory studies may be needed to reaffirm the findings in the current study also apply to a predominantly Caucasian population. Finally, the current study is limited in scope to men treated with radiation therapy for prostate cancer and therefore needs confirmatory studies before drawing definitive conclusions. Despite these limitations, the current study has demonstrated that patients with major psychiatric disorders who undergo treatment with radiation should expect similar biochemical, distant metastatic and survival outcomes as the general population with the same limited toxicity. Accordingly, this suggests that if barriers to health care are eliminated or reduced, clinical outcomes will be similar among those +Psy or −Psy.
In conclusion, in the current study of 469 men with prostate cancer who elected treatment with definitive radiation, there were no differences in any outcome or toxicity endpoint between +Psy or −Psy men. Further studies are needed to better delineate whether the reports of worse mortality in this population are cancer-specific outcomes or are in fact related to other comorbidities affected by their psychiatric illnesses.
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.