CYFRA 21-1 levels in cerebrospinal fluid as a putative therapeutic monitoring biomarker for patients with leptomeningeal carcinomatosis: A pilot study

Abstract

BACKGROUND:

To investigate the feasibility of cerebrospinal fluid (CSF) CYFRA 21-1 levels as a therapeutic monitoring biomarker in leptomeningeal carcinomatosis (LMC) patients undergoing ventriculo-lumbar perfusion (VLP) chemotherapy.

METHODS:

The levels of CYFRA 21-1 in 42 CSF samples from 15 LMC patients were analyzed using an electrochemiluminescence immunoassay. Samples were collected at individual time points during VLP chemotherapy. Therapeutic outcomes were measured as improvements in the Karnofsky Performance Status (KPS) score and decreasing intracranial pressure (ICP) as the main endpoint of VLP chemotherapy. Changes in CSF CYFRA 21-1 levels, protein levels, and cytology results were also investigated. We subsequently evaluated whether these changes were correlated with KPS score and ICP.

RESULTS:

The CSF CYFRA 21-1 levels at individual time points were associated with KPS score and ICP. The KPS scores ( $p=$ 0.007) and ICP ( $p=$ 0.018) of patients with high CSF CYFRA 21-1 levels were significantly different from those of patients with low CSF CYFRA 21-1 levels. By contrast, CSF protein levels and cytological responses were not significantly associated with KPS scores and ICP.

CONCLUSIONS:

CSF CYFRA 21-1 may have utility as a therapeutic monitoring biomarker to design personalized therapeutic strategies in LMC patients undergoing VLP chemotherapy.

Keywords

Leptomeningeal metastasis monitoring biomarker CSF CYFRA 21-1

1. Background

CYFRA 21-1 is a fragment of cytokeratin 19, a component of the structural proteins that maintain the stability of epithelial cells, and has been investigated as a serum biomarker in several cancers [1, 2, 3, 4]. Recent studies have reported evidence supporting CYFRA 21-1 in cerebrospinal fluid (CSF) as a diagnostic and prognostic indicator for leptomeningeal carcinomatosis (LMC), a multifocal involvement of malignant cells in the leptomeninges and/or CSF in metastatic solid tumors [5, 6, 7, 8, 9, 10, 11].

With current advances in understanding, and improvement in diagnostic tools and systemic cancer therapies, the survival of cancer patients has been prolonged and the incidence of LMC has also increased [11, 12, 13]. LMC was generally regarded to be a debilitating condition, leading to end-of-life care without an established optimal treatment. However, the development of new therapeutic methodologies, such as ventriculo-lumbar perfusion (VLP) chemotherapy, which decrease the disruption of CSF flow, have enabled us to expect longer survival of patients with LMC [14, 15, 16, 17]. For these reasons, evaluating treatment response in patients with LMC has become more important in decision making and designing personalized therapeutic strategies in clinical practice. Nevertheless, established biomarkers that can monitor the benefits of specific regimens in LMC patients, particularly in those involving new therapeutic options such as VLP chemotherapy are still unavailable.

In a previous study, we preliminarily reported the potential of CSF CYFRA 21-1 as a therapeutic monitoring tool for LMC patients undergoing VLP chemotherapy [9]. In the present study, we further evaluated the feasibility of CSF CYFRA 21-1 as a therapeutic monitoring biomarker through assessment at individual time points in additional LMC patients undergoing VLP chemotherapy.

2. Methods

LMC patients ( $n=$ 15) enrolled in an ongoing clinical trial investigating the optimal perfusion rate for VLP chemotherapy at the National Cancer Center between 2015 and 2018 were included. Patients were eligible for inclusion if CSF samples and adequate medical records for evaluation of therapeutic outcomes, including the Karnofsky Performance Status (KPS) scores, intracranial pressure (ICP), CSF protein levels, and cytological results were available. Two patients included in the authors’ previous study, for whom additional samples with a longer follow-up duration were available, were also enrolled in the present study(Fig. 1A and B in this study correspond to Fig. 3A and E in the previous study) [9]. Among patients in the present study, the most common primary tumor was lung cancer ( $n=$ 8), followed by breast cancer ( $n=$ 4), and ovarian cancer ( $n=$ 2), one of which had a tumor with unknown origin. Histologically, all patients presented with adenocarcinoma. LMC was diagnosed based on positive CSF cytology results and/or magnetic resonance imaging findings compatible with LMC. VLP chemotherapy was performed at a perfusion rate of 15 mL/h, and a daily methotrexate dose of 24 mg was administered. CSF CYFRA 21-1 levels were measured before and after commencing VLP chemotherapy at individual time points in patients with LMC. This study was approved by the Institutional Review Board of the National Cancer Center (NCC 2017-0135), and written informed consent was obtained from all patients.

Figure 1.

Analysis of therapeutic responses during ventriculo-lumbar perfusion (VLP) chemotherapy in patients with leptomeningeal carcinomatosis. The black squares represent the serial levels of cerebrospinal fluid (CSF) CYFRA 21-1, the gray circles represent the serial levels of intracranial pressure, and the linear bars above the graph represent VLP chemotherapy. The Karnofsky performance status (KPS) score and CSF cytology results are presented below the graphs. ( $+$ ) represents positive malignant cells in the CSF, ( $-$ ) represents negative malignant cells in the CSF, and ( $\pm$ ) represents atypical cells in the CSF.

Figure 1.

continued.

Figure 2.

Analysis of therapeutic responses during ventriculo-lumbar perfusion (VLP) chemotherapy in patients with leptomeningeal carcinomatosis. The black squares represent the serial levels of cerebrospinal fluid (CSF) protein, the gray circles represent the serial levels of intracranial pressure, and the linear bars above the graph represent VLP chemotherapy. The Karnofsky performance status (KPS) score and CSF cytology results are presented below the graphs. ( $+$ ) represents positive malignant cells in the CSF, ( $-$ ) represents negative malignant cells in the CSF, and ( $\pm$ ) represents atypical cells in the CSF.

Figure 2.

continued.

Table 1

Different level of CSF CYFRA 21-1 using cut-off value at the individual time points during VLP chemotherapy according to the clinical or biological status

	CSF CYFRA 21-1 $>$ 2.94 ng/ml ( $n=$ 28)	CSF CYFRA 21-1 $\leqslant$ 2.94 ng/ml ( $n=$ 14)	$p$ -value
Karnofsky performance status score $<$ 70	24	5
Karnofsky performance status score $\geqslant$ 70	4	9	0.0032
Intracranial pressure ( $\geqslant$ 200 mmH ${}_{2}$ O)	15	2
Intracranial pressure ( $<$ 200 mmH ${}_{2}$ O)	13	12	0.0204

Abbreviation: VLP, ventriculo-lumbar perfusion, NS; not significant.

Table 2

The comparison of therapeutic responses between groups with different level of CSF CYFRA 21-1 and CSF protein

	CSF CYFRA 21-1 $>$ 2.94 ng/ml ( $n=$ 28)	CSF CYFRA 21-1 $\leqslant$ 2.94 ng/ml ( $n=$ 14)	$p$ -value
Median Karnofsky performance status score	60	70	0.007
Median intracranial pressure (mmH ${}_{2}$ O)	205	135	0.018
	CSF protein $>$ 50 mg/dl ( $n=$ 15)	CSF CYFRA 21-1 $\leqslant$ 50 mg/dl ( $n=$ 27)	$p$ -value
Median Karnofsky performance status score	60	60	NS
Median intracranial pressure (mmH ${}_{2}$ O)	150	170	NS

Abbreviation: NS; not significant.

The concentrations of CSF CYFRA 21-1 were determined using a commercially available electrochemiluminescence immunoassay (Elecsys CYFRA 21-1, Roche Diagnostics, Penzberg, Germany) according to the manufacturer’s instructions [18]. CYFRA 21-1 levels were estimated from 42 preserved CSF samples that were collected from the lumbar subarachnoid space and stored at $-$ 80 ${}^{\circ}$ C. To minimize the dilution effect of VLP drainage, CSF CYFRA 21-1 levels measured promptly before the initiation of the subsequent cycle of VLP chemotherapy were analyzed. Samples for which CSF CYFRA 21-1 levels were higher than a previously determined prognostic cut-off value (2.94 ng/mL) [9] despite the dilution effect of VLP chemotherapy, were also analyzed. Measurement of CSF CYFRA 21-1 levels was independently performed by investigators blinded to the study endpoint.

Therapeutic outcomes were evaluated based on clinical improvement in the KPS score during treatment, and decreasing ICP as a main biological indicator of treatment response in LMC patients undergoing VLP chemotherapy [15, 16]. We analyzed positive ( $>$ 2.94 ng/ml) and negative ( $\leqslant$ 2.94 ng/ml) status of CSF CYFRA 21-1 levels using the cut-off value at individual time points during VLP chemotherapy, according to the different clinical (possible to care for self, KPS $\geqslant$ 70) and biological (increasing intracranial pressure $\geqslant$ 200 mmH ${}_{2}$ O) status. Additionally, we compared the therapeutic outcomes of groups with high and low CSF CYFRA 21-1 levels (cut-off, 2.94 ng/mL) [9], and CSF protein levels (cut-off, 50 mg/dL).

Fisher’s exact test was performed to compare different status of CSF CYFRA 21-1 levels at individual time points according to the clinical and biological status. Mann-Whitney test or Student’s $t$ -test was used to compare differences in continuous variables between the groups. In all analyses, $p<$ 0.05 were considered to be statistically significant.

3. Results

3.1 Demographics

Among the 15 patients with LMC, the male-to-female ratio was 1:2, and the mean ( $\pm$ SD) age at sampling was 50.7 $\pm$ 11.7 years. The mean follow-up duration of individual CSF CYFRA 21-1 level assessment was 27.9 $\pm$ 26.9 days.

3.2 Assessment of CSF CYFRA 21-1 levels during VLP chemotherapy: Association with clinical, biological, and cytological responses

Decreasing CSF CYFRA 21-1 levels was associated with improvement of KPS scores in all but 2 cases (Fig 1C and I), and with decreasing ICP in all but 5 cases (Fig. 1C, G, J, M and O). Analysis of CSF protein levels, estimated 1 to 3 days after each CSF CYFRA 21-1 assessment, revealed a modest association with KPS score or ICP (Fig. 2B–E, M and N). By contrast, cytological responses were randomly associated with KPS scores and ICP at the individual time points (Figs 1 and 2).

3.3 Comparison of KPS score and ICP between groups with different CSF CYFRA 21-1 and CSF protein levels

Table 1 shows the high ( $>$ 2.94 ng/ml) and low ( $\leqslant$ 2.94 ng/ml) levels of CSF CYFRA 21-1 levels at individual time points according to the clinical (KPS $\geqslant$ 70 (possible to care for self), $p=$ 0.0032) or biological (increased ICP $\geqslant$ 200 mmH ${}_{2}$ O, $p=$ 0.0204) status, and significant differences were observed.

Significant differences in the therapeutic responses defined by KPS (median: 60 vs. 70, $p=$ 0.007) and ICP (205 vs. 135 mmH ${}_{2}$ O, $p=$ 0.018) between patients with high and low CYFRA 21-1 levels were observed (Table 2). In contrast, these outcomes did not significantly differ between patients with high and low CSF protein levels.

4. Discussions

We observed that the CSF CYFRA 21-1 levels at individual time points were associated with therapeutic responses reflected in the KPS scores and ICP as a major end point of clinical trials involving LMC patients undergoing VLP chemotherapy [15, 16]. When we compared the KPS scores and ICP of patients with high and low CSF CYFRA 21-1 levels, significant differences were observed. On the contrary, neither association of CSF protein levels and therapeutic outcomes nor difference in therapeutic outcomes between groups with high and low CSF protein levels, showed statistical significance. Further, cytological responses were randomly associated with therapeutic responses. These results suggest that CSF CYFRA 21-1 may serve as a therapeutic monitoring biomarker, and may facilitate the design of individualized therapeutic strategies for LMC patients treated with VLP chemotherapy.

As a biomarker, CSF CYFRA 21-1 can provide objective quantitative data for therapeutic monitoring in LMC patients, whereas CSF cytology yields only qualitative information. Furthermore, CYFRA 21-1 levels can be easily measured using preserved CSF samples and smaller CSF volumes are required compared to those used in CSF cytology testing [18, 19]. Additionally, through additional analysis of CSF protein levels at slightly delayed time points of CSF CYFRA 21-1 assessment, it appears that CSF protein is less immediately sensitive to therapeutic responses than CSF CYFRA 21-1. Therefore, CSF CYFRA 21-1 is likely to be a more well-timed biomarker than CSF protein.

VLP chemotherapy is administered by a passage from the intra-ventricular reservoir to the lumbar drainage site [15]. CSF could be a more specific source of biomarkers in patients with LMC than other body fluids, such as blood; however it is not always suitable for longitudinal therapeutic monitoring due to relative limitations in accessibility. Nevertheless, in patients undergoing VLP chemotherapy, longitudinal CSF samples can be easily obtained from the intra-ventricular reservoir or lumbar drainage site. In contrast to our previous study, which included CSF samples from the lumbar subarachnoid or intraventricular space [9], only samples from the lumbar subarachnoid space were evaluated in this study. By removing the potential influence of sampling site/source in analyzing CYFRA 21-1 [20], more consistent results could be obtained.

In this study, the prognostic cut-off value (2.94 ng/mL) for CSF CYFRA 21-1 was based on overall survival as an outcome in a previous study [9]. The use of cut-off values could be more convenient in clinical practice than absolute values. Using this cut-off value, we observed significant difference in the therapeutic responses of patients with high and low CSF CYFRA 21-1 levels. In addition, the median KPS score of patients with low CSF CYFRA 21-1 levels was 70, and KPS score $\geqslant$ 70 has been previously shown to predict favorable clinical outcomes [11]. Future studies based on a larger cohort of LMC patients with diverse endpoints including overall survival for the optimal cut-off value would be helpful in implementing personalized therapeutic decision making.

Methodological limitations of the current study include the enrollment of LMC patients from a single referral center and its retrospective design, which may have resulted in unintentional selection bias and estimation of CSF CYFRA 21-1 levels at irregular time intervals. Addressing these issues warrant further prospective multicenter studies with longitudinal assessments of CSF CYFRA 21-1 levels at regular time points.

5. Conclusion

In conclusion, CSF CYFRA 21-1 appears to be a potential therapeutic monitoring biomarker as a complement to other clinical and biological parameters in patients with LMC undergoing VLP chemotherapy.

Footnotes

Acknowledgments

This study was supported by the National Cancer Center (1741580-1) and the National Research Foundation of Korea (NRF-2018R1A5A2023127).

Conflict of interest

SHS, KSY, GHS, YH report no disclosures. HJW has received grants from the National Cancer Center and National Research Foundation of Korea. KSH has lectured, consulted, and received honoraria from Bayer Schering Pharma, Biogen, Genzyme, Merck Serono, and UCB and received a grant from the National Research Foundation of Korea. Dr. Kim received a grant from the National Research Foundation of Korea; received consultancy/speaker fees from Alexion, Celltrion, Eisai, HanAll BioPharma, Merck Serono, Novartis, Sanofi Genzyme, Teva-Handok, and Viela Bio; serves on a steering committee for MedImmune/Viela Bio; is a co-editor for the Multiple Sclerosis Journal and an associated editor for the Journal of Clinical Neurology.

References

Barak

Goike

Panaretakis

and Einarsson

, Clinical utility of cytokeratins as tumor markers, Clin Biochem 37 (2004), 529–540.

Nisman

Lafair

Heching

Lyass

Baras

and Barak

, Evaluation of tissue polypeptide specific antigen, CYFRA 21-1 and carcinoembryonic antigen in non small cell lung carcinoma: Does the combined use of cytokeratin markers give an additional information? Cancer 10 (1998), 1850–1859.

Park

S.Y.

Lee

J.G.

Kim

et al., Preoperative serum CYFRA 21-1 level as a prognostic factor in surgically treated adenocarcinoma of lung, Lung Cancer 79 (2013), 156–160.

Boeck

Wittwer

Heinemann

et al., Cytokeratin 19-fragments (CYFRA 21-1) as a novel serum biomarker for response and survival in patients with advanced pancreatic cancer, Br J Cancer 108 (2013), 1684–1694.

Gauthier

Guilhaume

M.N.

Bidard

F.C.

et al., Survival of breast cancer patients with meningeal carcinomatosis, Ann Oncol 21 (2010), 2183–2187.

Comte

Jdid

Guilhaume

M.N.

et al., Survival of breast cancer patients with meningeal carcinomatosis treated by intrathecal thiotepa, J Neurooncol 115 (2013), 445–452.

Wang

Piao

Zhang

and Hao

, The concentration of CYFRA 21-1, NSE and CEA in cerebro-spinal fluid can be useful indicators for diagnosis of meningeal carcinomatosis of lung cancer, Cancer Biomarkers 13 (2013), 123–130.

Zhang

Tian

Shi

et al., Diagnostic value of CYFRA 21-1 in the cerebrospinal fluid for leptomeningeal metastasis, Dis Markers 2017 (2017), 2467870.

Hyun

J.W.

Park

J.H.

Kang

B.G.

et al., Diagnostic and prognostic values of cerebrospinal fluid CYFRA 21-1 in patients with leptomeningeal carcinomatosis, Oncotarget 8(32) (2017 Jun 8), 53326–53335.

10.

Chamberlain

M.C.

, Leptomeningeal metastasis, Curr Opin Oncol 22 (2010), 627–635.

11.

Hyun

J.W.

Jeong

I.H.

Joung

Cho

H.J.

Kim

S.H.

and Kim

H.J.

, Leptomeningeal metastasis: Clinical experience of 519 cases, Eur J Cancer 56 (2016), 107–114.

12.

Groves

M.D.

, New strategies in the management of leptomeningeal metastases, Arch Neurol 67 (2010), 305–312.

13.

Freilich

R.J.

Krol

and DeAngelis

L.M.

, Neuroimaging and cerebrospinal fluid cytology in the diagnosis of leptomeningeal metastasis, Ann Neurol 38 (1995), 51–57.

14.

Nakagawa

Fujita

Kubo

et al., Ventriculolumbar perfusion chemotherapy with methotrexate and cytosine arabinoside for meningeal carcinomatosis: A pilot study in 13 patients, Surg Neurol 45 (1996), 256–264.

15.

Gwak

H.S.

Lim

H.S.

Shin

S.H.

et al., Ventriculolumbar perfusion chemotherapy for the treatment of leptomeningeal carcinomatosis: A phase I study with pharmacokinetic data, Am J Clin Oncol 36 (2013), 491–499.

16.

Gwak

H.S.

Joo

Shin

S.H.

et al., Ventriculolumbar perfusion chemotherapy with methotrexate for treating leptomeningeal carcinomatosis: A phase II study, Oncologist 19 (2014), 1044–1045.

17.

Gwak

H.S.

Lee

S.H.

Park

W.S.

et al., Recent advancements of treatment for leptomeningeal carcinomatosis, J Korean Neurosurg Soc 58 (2015), 1–8.

18.

Forster

R.J.

Bertoncello

and Keyes

T.E.

, Electrogenerated chemiluminescence, Annu Rev Anal Chem 2 (2009), 359–385.

19.

Nabors

L.B.

Portnow

Ammirati

et al., Central nervous system cancers, version 1, J Natl Compr Canc Netw 13 (2015), 1191–1202.

20.

Gajjar

Fouladi

Walter

A.W.

et al., Comparison of lumbar and shunt cerebrospinal fluid specimens for cytologic detection of leptomeningeal disease in pediatric patients with brain tumors, J Clin Oncol 17 (1999), 1825–1828.