Low Detection Rate of H3K27M Mutations in Cerebrospinal Fluid Obtained from Lumbar Puncture in Newly Diagnosed Diffuse Midline Gliomas

Abstract

Recent studies have suggested the feasibility of detecting H3K27M mutations in the cerebrospinal fluid of diffuse midline glioma (DMG) patients. However, cerebrospinal fluid from patients in these studies were collected mainly during biopsy, ventriculo-peritoneal shunt procedures or postmortem. We assessed circulating tumor DNA (ctDNA) extracted from cerebrospinal fluid (CSF) and plasma in a series of 12 radiographically suspected and/or pathologically confirmed diffuse midline glioma patients and assessed for H3F3A K27M mutation using digital droplet PCR. In 10 patients, CSF was obtained by lumbar puncture at presentation. A definitive detection of H3F3A K27M mutation was achieved in only one case (10%); H3F3A K27M mutation was suspected in three other cases (30%). H3F3A K27M mutation was detected in two patients in CSF obtained by ventricular tap during a ventriculo-peritoneal shunt for obstructive hydrocephalus. Cases in which a definitive assessment was possible (definite H3F3A K27M or definite H3F3A wildtype) tended to be younger (median 7.5 years vs. 40.5 years; p = 0.07) and have a higher concentration of CSF protein (median 123 mg/dl vs. 27.5 mg/dl; p = 0.21) compared to nondefinite cases. Low proliferation and apoptotic rates seemed to be characteristics of DMG unfavorable for liquid biopsy. More advanced lesions with necrosis and evidence of dissemination were unlikely to be candidates for lumbar puncture due to the fear of exacerbating obstructive hydrocephalus. Methods to safely sample CSF and a more sensitive detection of ctDNA are necessary for reliable liquid biopsy of DMG at presentation.

Methodological Challenges of Digital PCR Detection of the Histone H3 K27M Somatic Variant in Cerebrospinal Fluid

“…This situation reflects lower levels of wild-type positivity observed in dPCR with lumbar CSF-derived templates. We encountered two lumbar CSF samples from H3 K27M-positive DMG patients, similar to those described by On et al [ 23 ]—with low cfDNA concentrations and single positive signals in dPCR charts. For one of them, ventricular CSF was also available, and the result was interpreted unambiguously ( Figure 6 ).…”

“…Despite the significantly reduced concentrations of cfDNA in CSF derived by lumbar puncture, Fujioka et al detected mutations in IDH1 , TERT , or H3 genes in six of seven high-grade glioma cases [ 22 ] . At the same time, On et al were able to unequivocally identify H3 K27M mutation in only 1 of 10 CSF samples obtained by lumbar puncture from newly diagnosed DMG patients; for three other cases in this cohort, the result was ambiguous, as only 1-2 mutant droplets were observed in ddPCR charts [ 23 ]. In our experience, concentrations of lumbar CSF-derived cfDNA (0–10, median 3 copies/µl) were also lower compared with the isolates obtained from ventricular CSF (0–15,170, median 94.5 copies/µl).…”

“…In addition to the difficulties in CSF sample collection and primary handling, their advanced molecular analysis is sophisticated. Modern strategies for analysis of somatic mutations in cfDNA are based on the use of targeted high-throughput sequencing [ 12 - 14 , 19 - 21 ] or dPCR [ 14 , 16 , 21 , 23 , 25 ]. Apart from being an established method for detecting extremely low allelic loads, dPCR provides the unique ability to quantify target DNA without external standards.…”

Liquid biopsy and tumor DNA/RNA detection in the cerebrospinal fluid of patients diagnosed with central nervous system glioma – A review article

“…[ 2 ] About 50– 80% of these tumors carry mutations in the genes encoding histone 3, more specifically HIST1H3B (H3.1K27M) and H3F3A (H3.3K27M). [ 9 , 15 , 23 , 26 , 35 ] The discovery of such mutations brings hope for the development of new targeted therapies such as Impridone ONCO201 and IDO1 enzyme inhibitor which in preclinical trials have demonstrated activity against mutated H3K27M gliomas. [ 35 ] Through liquid biopsy of tumor DNA in CSF, we could diagnose such mutations, define the appropriate target therapies, and monitor the therapeutic response with the option of multiple throughout the disease course.…”

“…[ 10 , 15 , 20 ] On the other hand, studies using lumbar puncture to collect CSF resulted in a lower sensitivity for detection of tumor DNA when compared to the levels detected in samples collected through ventricular shunts. [ 23 ] Given the variety of diagnostic methods (Sanger, ddPCR, and NGS), biomarkers and collection methods used in the studies of this review, the need to standardize the steps for obtaining samples, analysis, and interpretation of results in liquid biopsy is evident.…”