Comparative evaluation of the diagnostic and prognostic performance of CNSide™ versus standard cytology for leptomeningeal disease

Abstract Background This retrospective study compares the real-world performance of cerebrospinal fluid (CSF) CNSide™ versus cytology in leptomeningeal disease (LMD). Methods Consecutive patients with suspected LMD who underwent lumbar punctures for CSF cytology and CNSide™ from January 2020 to December 2022 were reviewed. LMD was classified by EANO criteria. Descriptive statistics, confusion matrix, Kaplan–Meier curves, and Cox proportional regression were used. Results Median age for 87 evaluable patients was 63 years (range: 23–93); 82 (94%) met EANO criteria for possible/probable/confirmed LMD (EANO/LMD). The commonest primary cancers were breast (36,44.0%) and lung (34,41.5%). Primary lung harbored actionable mutations in 18 (53.0%); primary breast expressed hormone receptors in 27 (75%), and HER2 amplification in 8 (22%). Uncontrolled systemic disease was detected in 35 (40%), while 25 (46%) received systemic therapy with medium/high CNS penetrance at LMD diagnosis. The median time from initial cancer to LMD diagnosis was 31 months (range: 13–73). LMD was confirmed by CSF cytology in 23/82 (28%), all identified by CNSide™. CNSide™ identified 13 additional cases (36/82, 43.9%), increasing diagnostic yield by 56.5%. Median overall survival (mOS) was 31 weeks (95%CI: 21–43), significantly worse for CNSide™ positive versus negative: 4.0 versus 16.0 weeks, respectively (HR = 0.50, P = .010). While survival since LMD diagnosis did not differ by histology, time to LMD diagnosis from initial cancer diagnosis was longer for breast (48.5 months, IQR: 30.0–87.5) versus lung (8 months, IQR:0.5–16.0) cohorts. mOS was longer for patients eligible for intrathecal chemotherapy (HR: 0.189, 95%CI: 0.053–0.672, P = .010). Conclusions This retrospective, real-world analysis of CNSide™ showed increased sensitivity versus cytology and provided clinically relevant molecular CSF analyses.

2][3] LMD occurs when cancer spreads to the leptomeninges or subarachnoid space by hematogenous spread, direct infiltration from solid brain metastases, perineural, or perivascular invasion, or seeding of cancer cells after neurosurgical interventions. 1,4,5Once seeded in the meninges, cancer cells can circulate along the meningeal and ependymal surfaces or via cerebrospinal fluid (CSF) flow, with a preference for settling in regions of slow CSF flow,

Comparative evaluation of the diagnostic and prognostic performance of CNSide™ versus standard cytology for leptomeningeal disease
][4] LMD diagnosis is confirmed by the detection of malignant cells in the CSF or suggested by clinical and/or neuroimaging findings in patients with metastatic disease. 2 The diagnosis is often quite challenging due to limited sensitivity with initial cytology (50-67%) and MRI (75%), as well as high variability of clinical presentation. 2,3Clinical symptomatology is nonspecific and includes headache (from meningeal irritation), cranial nerve palsies, altered mental status, bowel or bladder dysfunction, extremity numbness and weakness, and symptoms of elevated intracranial pressure from LMD-induced irregularities in CSF dynamics. 2,3Using joint European Association of Neuro-Oncology and European Society of Medical Oncology [EANO-ESMO] criteria, LMD diagnosis is defined as type I (confirmed malignant cells in cytology), and type II (lack of verification from cytology). 2 Neuroimaging findings further classify LMD into different categories, linear disease (type A), nodular disease (type B), both linear and nodular (type C), and no imaging findings or hydrocephalus (type D). 2 This classification broadens LMD definition as confirmed, probable, possible, or lack of evidence (Table 1), providing guidance on treatment choice with reasonable confidence for "confirmed and probable" cases versus additional work-up advised for "possible or lack of evidence" cases. 2 CSF cytology remains the diagnostic gold standard for LMD with the highest specificity, however with low sensitivity, as malignant cells are detected in only 50-67% of initial lumbar punctures, increasing to 80-90% only after 3 serial lumbar punctures. 2,3Although minimally invasive, lumbar puncture is associated with potential complications including postprocedural headache, cranial neuropathies, nerve root irritation, low back pain, stylet-associated problems, infections, bleeding, and cerebral and spinal herniation. 6In addition to these potential effects, procedural time and patient reluctance further limit the feasibility of completing multiple spinal taps in the outpatient setting. 7urthermore, standardization of collection protocols are needed as inappropriate sampling such as limited quantity of CSF retrieved, and inaccurate technique can lead to incomplete analysis reducing the amount of accurate information obtained. 7urrently, the available diagnostic modalities for LMD do not provide reliable information on disease burden or (early) response to treatment. 3Recent molecular technological advances provide informative analysis of biofluids with the liquid biopsy, an alternative to tissue biopsy that involves the sampling of body fluids, such as CSF, for molecular components released from cells. 8,9Cell-free DNA (cfDNA) are short-fragmented DNA molecules in the plasma, and when released from tumor cells by apoptosis or necrosis, are termed circulating tumor DNA (ctDNA). 3,10These molecular tools provide the opportunity to identify specific and potentially actionable therapeutic targets and aid in early diagnosis, assessing treatment response, and monitoring for disease recurrence. 3,8,11Tumor-specific genomic abnormalities can be detected in ctDNA using next-generation sequencing (NGS) techniques or polymerase chain reaction (PCR)-based options, and have demonstrated a strong concordance with the genomic profile of malignant tissue, making ctDNA a potentially powerful biomarker. 3,11iven the poor sensitivity of standard cytology in diagnosing LMD, the primary aim of this study was to compare the real-world diagnostic and prognostic yield of CNSide TM assay (Biocept Inc.) versus standard cytology.The CNSide TM assay is a novel combination of cell-based and cell-free assays to provide quantitative analysis utilizing proprietary technology. 12,13Given the innovative approaches in genomic testing, such as using cfDNA in the CSF, this retrospective analysis also hoped to gain further understanding of prognostic features in patients with LMD by analyzing tumor cell (TC) count, cell density (TC/mL), and molecular features of circulating tumor cells (CTCs) in the CSF assays.The potential to diagnose and characterize CNS disease, replace, or complement ambiguous imaging studies, and monitor treatment response in a minimally invasive manner provide exciting opportunities for LMD detection and treatment.

Population
Consecutive patients at Miami Cancer Institute or Baptist Health Hospital, part of Baptist Health South Florida with suspected LMD from January 2020 to December 2022 who underwent lumbar punctures for CSF cytology and CNSide™ analyses were evaluated.LMD diagnosis was defined by EANO-ESMO criteria. 2As the study was retrospective in nature, the requirement for informed consent was waived by the institutional IRB review.

Importance of the Study
Leptomeningeal disease (LMD) is a rare, but highly morbid and fatal complication affecting 5-10% of cancer patients.CSF cytology still remains the diagnostic gold standard with the highest specificity, however, malignant cells are detected in only 50-67% of initial lumbar punctures.LMD diagnosis is made radiographically, clinically, or by cytology of cerebrospinal fluid (CSF) based on EANO criteria, defined as definite, probable, possible, or no evidence of LMD.In a single-institution series, CNSide™ increased the diagnostic yield by 56.5%.CNSide™ also showed increased sensitivity relative to standard cytology and increased specificity relative to EANO criteria and provided clinically relevant, -cell-based molecular and -cell-free DNA analyses.

CSF Assays
CSF cytology was analyzed as per standard practice.CSF tumor cells were also captured and characterized using CNSide TM assay.CSF collected at the time of lumbar puncture was placed into CNSide TM collection tube stabilized at ambient temperature.In the CLIA laboratory CSF-TCs are isolated using a proprietary mouse anti-human antibody cocktail with secondary biotinylated anti-mouse Ig. 13 All CSF cells are then captured in a streptavidin-coated micro-fluidic channel and fluorescently labeled with DAPI, streptavidin, keratin, and CD45 to discriminate nucleated CSF-TCs that are SA + and cytokeratin+, but CD45 negative. 13CSF-TCs are digitally imaged, localized and counted with specific X-Y coordinates in the channel that allow further characterization with biomarker analysis, including fluorescence in situ hybridization (FISH) for rearrangements, copy number variations, and immunocytochemistry (IHC) for proteins and enumeration. 12,13For cfDNA, once in the collecting tube, DNA is isolated for NGS. 12

Statistical Analysis
All statistical analyses were done using SAS 9.4 (SAS Inc.).Initially, we compared EANO criteria (confirmed, probable, possible, and lack of evidence) by both cytology and CNSide™ to identify their distribution.Test characteristics, such as sensitivity, specificity, positive predictive value, and negative predictive value, were estimated for CNSide™ (cytology as gold standard) and cytology (CNSide™ as gold standard).Then we assessed the concordance between receptor identification by Biocept and primary cancer samples, then stratified by lung cancer and breast cancer using either Chi-squared test or Fisher's exact test.We compared Biocept CSF volume between CNSide™ positive and negative groups using Mann-Whitney U test.We created Kaplan-Meier curves to compare overall survival between EANO-positive (confirmed, probable, possible) versus negative (lack of evidence), CNSide™ positive versus negative, as well as cytology positive and negative.We also compared overall survival between three groups: CNSide™ positive and cytology negative versus CNSide™ and cytology negative versus CNSide™ and cytology positive.We also compared overall survival between CNSide™ positive breast cancer and lung cancer patients.Finally, Cox proportional regression was performed to assess the association between cell density by CNSide™ and survival.All tests were two-sided and statistical significance was set at P < .05.
Commonest primary histologies in the EANO-positive cohort were breast (n = 36, 43.9%) and lung (n = 34, 41.5%).Among primary breast cancers, 30 (83.3%) expressed ER, 22 (61.1%)expressed PR, and in 8 (22.2%) had HER2 alterations.Among primary lung cancers, 17 (50.0%)cases harbored actionable molecular alterations (we should specify which ones, here).Table 4 shows the concordance between receptor status by CNSide™ versus primary tissue analyses for the whole cohort as well as lung versus breast cancer.Of note, CNSide™ detected a gain in KRAS (2, 5.6%) and EGFR (3, 8.3%) compared to the primary lung cancer biopsy, while a gain in ER (9, 23.1%), PR (3, 7.7%), and HER2 (2, 5.1%) and a loss of HER2 (3, 7.7.%)were noted compared to the primary breast cancer biopsy.Subgroup analyses were limited by sample size, but molecular profiles for all histologies and survival curves for the breast cases among available CNS metastasis surgical samples and the corresponding CSF CNSide TM molecular subgroups in our cohort are listed in Supplementary Tables 1 and 2.

Discussion
The incidence and prevalence of LMD have increased over past decades due to increased detection bias (particularly improved diagnostic imaging) as well as advances in systemic therapies that have resulted in increased survival (but often lack maximal or sufficient CNS and especially CSF penetrance). 5,14,15Breast cancer is the most common histology in patients with LMD, with lobular subtype and  triple-negative the most common risk factors. 5Prior surgical resection for brain metastases is associated with increased risk of leptomeningeal involvement, especially within the posterior fossa and resulting from a surgical spillover or drop metastasis of malignant cells into the leptomeninges induced by iatrogenic disruption of the blood-brain barrier. 5,15Early and accurate diagnosis of LMD, facilitated by advanced CSF analyses like CNSide TM , impacts diagnostic accuracy, facilitates proper prognostic estimates and discussions, and guides proper multimodal care. 15Wijetunga et al., in their proton craniospinal irradiation (pCSI) phase II trial cohort, showed that higher CSF ctDNA mean variant allele frequency (VAF) before pCSI (VAFpre) had worse OS (6 months for VAFpre ≥ 0.32 vs. 9 months for VAFpre < 0.32; P = .05).Similarly, increased VAF after pCSI portended worse survival (6 vs. 18 months; P = .008).Our cohort showed no similar correlation with pretreatment cell count density but warrants further validation in a larger prospective study.Serial CSF CTC and other assays also provide real-time and accurate response assessments that may in the future guide treatment changes and cessation by monitoring tumor dynamics and detection of molecular changes induced by treatment.
The increased utilization of blood-based liquid biopsies, specifically cfDNA, for molecular analysis in metastatic patients has served as an adjunct to tissue biopsies, and in some cases replaced repeat tissue biopsy altogether in patients who have declined surgical biopsy or deemed nonsurgical due to medical comorbidities.These liquid biopsies serve to monitor patients with known actionable molecular alterations not responding to targeted therapy and/or to further characterize concordance with primary genomic profiles.Tumor-specific genomic abnormalities can be detected in ctDNA using NGS techniques or PCRbased options, and have demonstrated a strong concordance with the genomic profile of malignant tissue, making ctDNA a potentially powerful biomarker. 3,11In a large-scale prospective concordance study of plasma and tissue NGS, Sugimoto et al. showed oncogenic alteration for nonsmall cell lung cancer was detected by plasma cfDNA sequencing in approximately 43% of patients and tissue assay in approximately 50% of patients.The concordance of plasma cfDNA sequencing compared with tissue DNA and RNA assays was 77% (EGFR, 78%; KRAS, 75%; BRAF, 85%; and HER2, 72%) and 47% (ALK, 46%; RET, 57%; ROS1, 18%; and MET, 66%), respectively. 16Of note, positive oncogenic drivers positive in plasma cfDNA and negative in tissue may result from unsuccessful genomic analysis from poor-quality tissue samples, while oncogenic drivers negative in plasma cfDNA and positive in tissue may result from the low sensitivity of cfDNA analysis or response to therapy. 16In patients with CNS disease, near-universal genomic divergence is seen across primary tumors and brain metastases with more than 50% of brain metastases harboring clinically actionable mutations not detected in the primary tumor. 179][20] Discordance may partly result from patients with LMD receiving prior systemic therapy and/or radiation therapy causing selective     A diagnostic lumbar puncture (LP) was performed at the bedside or occasionally via imaging guidance in the outpatient or occasionally inpatient settings.CSF was sent to cytology (volume 1-3 ml) and simultaneously shipped for CNSide testing (volume provided 2-10 ml, requested 5-8 ml).Upon receipt at the processing center, the CSF sample is divided for CTC isolation using a cancer-specific antibody cocktail.The CTC count and density preliminary report generated by the enumeration assay, followed by biomarker analysis including ICC for protein expression, FISH for rearrangements, and copy number variations.Simultaneously DNA isolation is tested for mutations by NGS methods.
penetrance of systemic therapy at the time of LMD diagnosis (H).We also compared overall survival overall survival from the time of primary cancer diagnosis (E) and from the time of leptomeningeal diagnosis (F).Multiple comparison (Figure I) shows significant differences in median OS between CNSide positive and cytology negative (1) versus CNSide and cytology negative (2) cohorts (P = .007),but no significant difference versus the CNSide and cytology positive (3) cohort.

Appel et al.: Comparative evaluation of the diagnostic and prognostic performance of CNSide™
pressures in the CSF cancer milieu. 18In our leptomeningeal cohort, CNSide™ detected a gain in KRAS (2, 5.6%) and EGFR (3, 8.3%) compared to the primary lung cancer profile, while a gain in ER (9, 23.1%), PR (3, 7.7%), and HER2 (2, 5.1%) and a loss of HER2 (3, 7.7.%)were noted compared to the primary breast cancer profile.Discordance was more common for primary breast cancer and overall showed gains in alterations rather than loss of expression.Potentially clinically informative alterations in the CSF can impact treatment decisions when selecting systemic therapy, however, most genomic alterations seen in the CNS cannot be targeted with currently available CNSpenetrant targeted therapies. 17,23SF-based liquid biopsies offer a minimally invasive surrogate to the higher-risk neurosurgery required for brain and spine tissue biopsies and, as our study shows, can even increase the yield of standard CSF and MRI diagnostic tools with respect to leptomeningeal metastases.The increased sensitivity, retained specificity, quantitative circulating tumor cell analysis, and extensive genomic profiling provided by CNSide TM typically required up to 8-10 mL of CSF, about 6-7% of the estimated 130-150 ml total volume of CSF in adults, and resulted in 1-3 weeks typically.Of note, the CNSide TM assay provided a significantly higher detection rate compared to standard CSF cytology, increasing the diagnosis of LMD by 56.5%.The CNSide TM assay provided extensive, tumor-specific, molecular data not available by standard CSF tests.Alternative paradigms exist for CSF qualitative and molecular profiling.Genomic Testing Cooperative (GTC) developed an approach to isolate cell-free total nucleic acid (cfDNA) followed by targeted NGS to sequence cell-free RNA (cfRNA) and cfDNA as novel liquid biopsy approach for both blood and CSF samples.24 GTC combines cfRNA and cfDNA NGS to evaluate mutations, fusion genes, and chromosomal structural abnormalities in liquid samples.24 cfRNA proved overall more sensitive than cfDNA in detecting mutations and, thus, reliable in detecting fusion genes and cfDNA is reliable in detecting chromosomal gains and losses.24 The liquid transcriptome, thus, can be used for diagnosis and staging, selecting therapy, predicting prognosis, and monitoring disease.Improved detection rates by these liquid biopsy strategies increase the certainty of LMD diagnosis, potentially enhancing prognostication and improving patient care by allowing for timely and appropriate radiotherapy, systemic therapy, and/or intrathecal chemotherapy selection.Optimizing multimodal and aggressive treatment for LMD diagnosis, especially earlier in the disease process before neurological disability or CSF flow abnormalities including hydrocephalus limit treatment options and prognosis, may improve survival for this rare but highly morbid leptomeningeal cancer diagnosis.[25][26][27]

Conclusions
This retrospective, single-institution, and real-world analysis of CNSide™ showed increased sensitivity versus cytology and provided clinically relevant, cell-based molecular CSF analyses.The CNSide™ assay readily identified tumor cells circulating in the CSF, increasing the accuracy and potentially hastening LMD diagnosis, while also providing quantitative data on tumor burden and treatment response.The enumeration of CTC did not provide prognostic data in this cohort and future analyses with larger cohorts are necessary for prognostic estimates to guide treatment decisions.In the era of precision medicine, detection of actionable genomic alterations by using assays such as CNSide TM and development of advanced trial designs and improved targeted intracranial penetrating therapies will help guide multimodal care for patients with otherwise dismal prognosis.Continued advancements in our understanding of LMD may provide a brighter outlook for these patients with poor prognosis, and it is imperative that neurocognitive decline and the patient's welfare are at the forefront of any treatment decision.

5 AFigure 1 .
Figure 1.CNSideTM sensitivity by CSF volume and EANO Criteria.(A) A box plot of CSF volume (ml) was not statistically different between CNSide positive or negative status.(B) Color map of cytology and CNSide positive or CNSide positive only among the suspected cases of leptomeningeal metatases categorized by EANO criteria.
Appel et al.: Comparative evaluation of the diagnostic and prognostic performance of CNSide™Overall Survival for EANO Positive Cases 1

Overall
Positive and Cytology Neg CNSide and Cytology 2: CNSide and Cytology Negative 3: CNSide and Cytology Positive Survival of CNSide TM Positive vs. CNSide TM Negative Cases Overall Survival by Systemic Therapy with Medium/High vs. Low CNS Penetrance Product-Limit Survival Estimates Product-Limit Survival Estimates with Number of Subjects at Risk and 95% Confidence Limits with Number of Subjects at Risk and 95% Confidence Limits Overall Survival of CNSide TM & Cytology Positive vs. CNSide TM Positive & Cytology Negative vs. CNSide TM & Cytology Negative Subsets EANO = European Association of Neuro-Oncology; CNS = Central Nervous System Product-Limit Survival Estimates with Number of Subjects at Risk and 95% Confidence Limits

Figure 2 .Figure 3 .
Figure 2. Overall survival by Kaplan-Meier.Kaplan-Meier overall survival curves for various subgroups and comparisons in our leptomeningeal metastases (LMD) cohort, including overall survival for the EANO-positive (confirmed, probable, possible) cohort (A), by status of systemic disease (B), cytology positive and negative (C), CNSide™ positive versus negative (3D), candidacy for intrathecal chemotherapy (G), and CNS

Table 1 .
EANO Diagnostic Criteria for Leptomeningeal Metastasis Disease (LMD) (IQR) from initial cancer diagnosis to LMD was 31 (21-43) months, statistically longer for primary breast cancer cohorts compared to lung cancer cohorts: 48.5 (30.0-87.5 1sed with permission and modified from Le Rhun et al (2017)1.EANO: European Association of Neuro-Oncology.Type A: LMD with typical linear MRI abnormalities; type B: LMD with nodular disease; type C: LMD with both linear and nodular disease; type D: LMD without MRI abnormalities (except hydrocephalus).CSF, cerebrospinal fluid; LMD, leptomeningeal metastasis disease; MRI, magnetic resonance imaging; NA, not applicable.a Requires a history of cancer with a reasonable risk of LM and consideration of alternative diagnoses.b Including in patients with a history of cancer.time

Table 2 .
Demographics at LMD Diagnosis

Table 4 .
Concordance Between Receptor Identification by Biocept and Tissue for Overall Sample, Lung Cancer, and Breast Cancer