Evaluating the cerebrospinal fluid ctDNA detection by next-generation sequencing in the diagnosis of meningeal Carcinomatosis

Meningeal carcinomatosis (MC), also called neoplastic meningitis (NM) is the most severe form of brain metastasis and causes significant morbidity and mortality [1]. The early diagnosis and timely treatment are likely to have the greatest impact on improving outcome. However, MC is easily to be missed in diagnosis and easily misdiagnosed because of diverse clinical manifestations and the lack of sensitive and specific diagnostic tools, which has presented difficulties for early treatment of patients with MC [2].

Currently, the diagnosis of MC is routinely confirmed on the basis of clinical signs and symptoms, cerebrospinal fluid (CSF) cytology, and/or neuroimaging (contrast-enhanced brain MRI and/or CT) findings [2]. The presence of neoplastic cells in the CSF is the most useful finding to confirm the diagnosis and CSF cytology remains the gold diagnostic standard, but 25–30% of suspected clinical cases with MC diagnosed based on clinical picture and neuroimaging findings will not be confirmed by this method [3, 4]. Although the positive rate of CSF cytology for neoplastic cells will increase with the number of punctures, the false negative rate is still high, due to the similarity between tumor cells and ependymal cells, contamination by blood brought about by multiple punctures, difficulty of the distinction lymphoma cells caused by viral infections of central nervous system (CNS), few tumor cells in the collected specimen and inadequate preparation of the sample [5]. Measurement of tumor markers in the CSF may be convenient and of value in the adjunctive diagnosis of MC but lack sensitivity and specificity [6‐9]. Amplification of tumor specific gene sequences by polymerase chain reaction may be used to detect minimal quantities of neoplastic cells in the CSF, which can be used when the gene mutations in the tumor are known but many cases of MC can be the first presentation of an as yet undetected primary cancer [10]. Contrast-enhanced brain MRI and/or CT is the technique of choice to evaluate patients with suspected MC but still has a 30% incidence of false-negative results [11]. Therefore, it is important to evaluate an alternative method for diagnosis of MC especially for cases with persistently negative CSF cytology and/or persistently negative neuroimaging results.

Circulating tumor DNA (ctDNA) consists of short, double stranded DNA fragments that are released by tumors [12]. The ctDNA implies the entire spectrum of tumor genome aberrations. Current approaches for detection of tumor genome aberrations in ctDNA include PCR-based methods targeting specific mutations and next-generation sequencing (NGS)-based options allowing detection of all possible aberrations in DNA. Improvements in read-length, sequence quality, and throughput allowed NGS to become a more promising method for quantifying ctDNA than PCR-based methods [13]. The tumor-derived ctDNA of plasma has been applied to detect multiple different types of cancer [14, 15]. However, plasma ctDNA from tumors confined to central nervous system was infrequently detectable given that physical obstacles such the blood-brain barrier could prevent ctDNA from entering the blood circulation [14, 16, 17]. Many studies have shown that CSF ctDNA could be an important method of liquid biopsy in patients with CNS cancers [17, 18]. However, only a single tumor type or single primary tumor type was included in these studies. Given that MC is involvement of the leptomeninges by metastatic tumors and can be observed in various kinds of solid tumors, we hope to evaluate the amounts of ctDNA among different primary tumor types and evaluate the clinical value of CSF ctDNA as liquid biopsy medium in the diagnosis of MC. Although the liquid biopsy approach has been shown to be promising, the sensitivity of CSF ctDNA with respect to conventional diagnosis of CSF cytology and neuroimaging findings have not been evaluated in patients with MC. Here, we evaluated the sensitivity of CSF ctDNA with NGS technology as liquid biopsy medium in the diagnosis of patients with MC.

In the present study, we compared the sensitivity of three methods including CSF cytology, neuroimaging findings and CSF ctDNA in the diagnosis of MC. The study revealed that CSF ctDNA had a higher sensitivity than the CSF cytology and neuroimaging findings. Most studies of ctDNA published to date have demonstrated the importance of plasma ctDNA as a liquid biopsy medium for various kinds of cancer given that benign tumors and non-neoplastic conditions do not generally give rise to ctDNA [23]. Similarly, CSF ctDNA may be useful in complementing the diagnosis of MC especially for cases with persistently negative CSF cytology and/or negative neuroimaging findings.

Currently, the presence of neoplastic cells in the CSF is the most useful criteria to confirm the diagnosis of MC and CSF cytology remains the gold diagnostic standard although the false negative rate of CSF cytology is still high. In our study, we found the sensitivity of CSF ctDNA is higher than that of CSF cytology. Previous study [14] had presented that ctDNA was often present in patients without detectable circulating tumor cells. Contrast-enhanced brain MRI is the technique of choice to evaluate patients with suspected MC but still has a 30% incidence of false-negative results [11]. The sensitivity of the neuroimaging was 63% compared to that of CSF-derived ctDNA by next-generation sequencing in our study. Neuroimaging studies are often noninformative or slow to reflect progression. Repeated neuroimaging also subjects patients to radiation, whereas monitoring ctDNA is noninvasive. However, other studies demonstrate that a cancer containing ~ 50 million malignant (rather than benign) cells releases sufficient DNA for detection in the circulation [24]. A cancer of this size is well below that required for definitive imaging at present. Measurement of tumor markers in the CSF may be convenient and of value in the adjunctive diagnosis of MC but lack sensitivity and specificity [6‐9]. Unlike tumor markers such as CA19–9 or CEA, which are expressed in normal cells as well as in neoplastic cells, gene mutations of a clonal nature are only found in neoplasms.

In cancer patients, ctDNA are thought to be released in plasma as a result of tumor cells apoptosis and/or necrosis [14, 25]. ctDNA has played an important role in monitoring disease status of advanced cancer patients as a promising blood-based biomarker [26]. Many studies showed that ctDNA analysis can be utilized in early diagnosis of human malignancies including pancreatic, renal, advanced ovarian, colorectal, bladder, gastroesophageal, breast, melanoma, hepatocellular, head and neck cancers [14, 19, 27‐29]. Previous study [14] showed that less than 10% of patients with gliomas harbored detectable ctDNA in the plasma while Yuxuan Wang et al. [18] demonstrated they identified detectable levels of CSF ctDNA in 74% of patients with primary tumors of the brain or spinal cord whereas no ctDNA was detected in patients whose tumors were not directly adjacent to a CSF reservoir. We attribute the 100% detection rate of CSF-derived ctDNA in our study to the reason that MC disseminates over the leptomeningeal surface, neoplastic cells shed into CSF and CSF is in direct contact with neoplastic cells and meninges while owing to physical obstacles such the blood–brain barrier, CSF ctDNA is unable to circulate fully within the blood system, resulting in a limited amount of ctDNA from CNS being released to plasma circulation.

Elena et al. [19] used next-generation sequencing to reveal somatic alterations in tumor-derived DNA from CSF in patients with CNS metastases mainly brain parenchyma metastasis while the positive rate is not 100 %. Our study showed a good result due to the patients who were recruited in this study were all definite MC cases. Tumor cells of MC cases were circulating in the CSF, which led to be easier to be caught. Consequently, this technique is more suitable for the diagnosis of MC than brain parenchyma metastasis.

As per the analysis for concordance of EGFR activating mutations between primary lung adenocarcinomas and MC, on the basis of patients’ CSF samples, the results showed that, EGFR activating mutations in CSF samples were consistent with those in primary adenocarcinomas. The showing of T790 M in two CSF samples, is possibly attributed to the fact that both two CSF samples were sequentially collected during the process of TKI therapy.

Our study find DNA mutation in CSF of patients with MC at 100% of our cohort, and it may give additional information to diagnose MC with negative CSF cytology. Furthermore, the type of tumor-associated gene mutations can provide many clues to the primary tumor type especially for patients whose primary tumor wouldn’t be found in spite of using various inspection methods. Additionally, looking for and knowing the primary tumor is the rather important diagnostic dependency.

However, our current study does have some limitations; for example, the cohort has had relative small numbers of patients and we used a smaller 143 gene panel. Though it was the most comprehensive gene panel at that time, including hot-spot mutation, copy number variants, gene fusion, and gene therapy information, it could not detect all possible aberrations in DNA. Thus, future study with larger size of samples from other departments and multiple institutions and using larger gene panel could help us to solve these issues.

This study suggests a higher sensitivity of CSF ctDNA than those of CSF cytology and neuroimaging findings. We find cancer-associated gene mutations in ctDNA from CSF of patients with MC at 100% of our cohort, and utilizing CSF ctDNA as liquid biopsy technology based on the detection of cancer-associated gene mutations may give additional information to diagnose MC with negative CSF cytology and/or negative neuroimaging findings. At the same time, cancer-associated gene mutations provide a reliable theory for MC targeted therapy.