From NGS panels to tumor mutational burden - how genetic profiling is reshaping treatment selection.
For most of the twentieth century, cancer treatment was organized around tissue of origin. Breast cancer patients received breast cancer protocols. Lung cancer patients received lung cancer protocols. The primary determinant of treatment selection was where in the body the tumor arose, not what molecular machinery was driving it.
That model is functionally obsolete. It has been replaced - not uniformly and not without resistance, but irreversibly - by genomics-guided treatment selection. The reason is straightforward: two patients with histologically identical tumors can have dramatically different molecular profiles, and those differences predict response to specific therapies with a precision that tissue-of-origin categorization cannot match.
The BRAF V600E mutation appears in melanoma, colorectal cancer, thyroid cancer, and non-small cell lung cancer. In all four cancer types, the presence of this variant predicts response to BRAF inhibitors like vemurafenib and dabrafenib. The tumor's location in the body is secondary to its molecular signature as a driver of treatment selection. This is the core logic of precision oncology, and genomic profiling is how you access it.
Next-generation sequencing (NGS) has transformed the practical feasibility of genomic profiling in clinical oncology. Where first-generation sequencing methods required weeks and significant cost to analyze individual genes, targeted NGS panels can now sequence 300-500 cancer-relevant genes simultaneously from a single tissue biopsy specimen, returning results in 10-14 days at a cost that most commercial insurers now cover for advanced-stage diagnoses.
Major panels - Foundation One CDx, Tempus xT, Caris Molecular Intelligence - report variants across multiple categories: somatic mutations (single-nucleotide variants and insertions/deletions), copy number alterations, gene fusions, and tumor mutational burden (TMB). Each category has distinct clinical implications.
Somatic mutations in actionable oncogenes like EGFR, ALK, ROS1, MET, and ERBB2 directly predict which targeted therapies are likely to produce response. Copy number amplifications in ERBB2 (HER2) drive treatment selection for a growing list of tumor types beyond breast cancer, following the 2022 FDA approval of trastuzumab deruxtecan for HER2-positive solid tumors regardless of histology. Gene fusions involving NTRK1/2/3 predict response to larotrectinib, another tumor-agnostic approval.
What NGS panels miss is equally important to understand. They do not measure protein expression levels, which matter for some biomarkers like PD-L1 that require immunohistochemistry. They do not capture epigenetic modifications that drive some cancer subtypes. And targeted panels by definition miss variants outside their covered gene set - a limitation that whole-exome and whole-genome sequencing approaches can address but at significantly higher cost and longer turnaround time.
Tumor mutational burden (TMB) has become one of the most discussed genomic biomarkers in oncology, following the 2020 FDA approval of pembrolizumab for TMB-high solid tumors. Understanding what TMB actually measures - and its limitations as a predictive biomarker - is important for oncologists incorporating it into treatment decisions.
TMB is a count of the number of somatic mutations per megabase of sequenced genome. Tumors with high TMB have accumulated more random mutations, which typically produces more neoantigens - aberrant protein fragments displayed on tumor cell surfaces that immune cells can recognize and attack. The hypothesis is that high-TMB tumors are more visible to the immune system and therefore more likely to respond to immune checkpoint inhibitors that remove suppressive signals on T cells.
The clinical reality is more nuanced. The initial pembrolizumab approval for TMB-high tumors was based on the KEYNOTE-158 basket trial, which showed response rates of 29% in TMB-high patients versus 6% in TMB-low patients. That is a real and meaningful difference. But it also means that 71% of TMB-high patients did not respond - a reminder that TMB is a probabilistic predictor, not a binary on/off switch for immunotherapy response.
The optimal TMB threshold for treatment decisions varies by tumor type. The 10 mutations/megabase cutoff used in the FDA approval is a population-level approximation. Colorectal cancers with POLE mutations can have TMB exceeding 100 mut/Mb and respond reliably to checkpoint inhibitors. Non-small cell lung cancers at the 10 mut/Mb threshold show more heterogeneous responses. Context matters, and TMB is most useful as one input among several rather than as a standalone decision criterion.
Somatic genomic profiling - testing the tumor - is now standard care for most advanced cancers. Germline testing - testing the patient's inherited genetic constitution - is following a similar trajectory but with different indications and a different clinical calculus.
The distinction matters because germline pathogenic variants have implications beyond the patient's own treatment. A germline BRCA2 variant in a patient with pancreatic cancer means that the patient's first-degree relatives have a 50% probability of carrying the same variant - and with it, elevated lifetime risk for breast, ovarian, and pancreatic cancers. Identifying and counseling those relatives is a clinical obligation, not just an informational footnote.
Current NCCN guidelines recommend germline testing for pancreatic cancer (all patients), ovarian cancer (all patients), colorectal cancer (Lynch syndrome evaluation in all patients under 70 and selected older patients), and prostate cancer (patients with high-risk or metastatic disease and a family history). The coverage of germline testing in treatment workflows is expanding, but implementation remains inconsistent across health systems.
Pegasi incorporates germline variant data from partner laboratory systems into our multi-modal diagnostic framework, flagging patients with pathogenic germline variants for cascade testing recommendations. For patients where germline testing has not yet been ordered but clinical criteria warrant it, the platform generates a care gap alert. Closing the germline testing gap is a population health intervention as much as an individual treatment decision.
One of the underappreciated aspects of genomics-guided treatment is that the relevant molecular profile changes over time. Tumors under therapeutic pressure evolve. EGFR-mutant lung cancers treated with first-generation EGFR inhibitors (erlotinib, gefitinib) commonly develop resistance through the T790M secondary mutation, which predicts response to third-generation osimertinib. Monitoring for the emergence of this resistance mutation - either through repeat tissue biopsy or through liquid biopsy - has become standard in EGFR-positive NSCLC management.
The liquid biopsy approach to serial profiling offers an important practical advantage: it can be performed with a blood draw rather than a repeat tissue procedure, allowing more frequent monitoring without the procedural risk and patient burden of serial biopsies. ctDNA assays can detect the emergence of resistance mutations as early as 4-6 weeks before radiographic evidence of disease progression on imaging - providing a decision window that tissue-based monitoring cannot match.
Integrating serial molecular profiling data into a longitudinal diagnostic model is one of the areas where Pegasi's architecture provides the most clinical value. Each new data point - whether a baseline NGS panel, a follow-up ctDNA assay, or a repeat CT scan - is incorporated into the patient's evolving model, updating the risk stratification and surfacing treatment implications in real time rather than requiring a fresh specialist consultation to interpret each result in isolation.
Here is the uncomfortable reality that most genomics vendors would prefer not to discuss: a substantial proportion of genomic data that is ordered and returned is never fully acted upon. A 2022 study in JCO Precision Oncology found that among patients with metastatic solid tumors who received NGS testing, only 41% had a documented discussion of the genomic results in their medical record within 90 days of report return. For patients at community oncology practices rather than academic medical centers, the proportion was lower.
The reasons are multiple. Oncologists at community practices may not have the time or specialized training to interpret a 40-page genomic report. The report may return after the treatment decision window. The relevant targeted therapy may not be available at that institution. Or the results may simply be filed and forgotten in the EHR without triggering any alert that the ordering physician would notice.
Pegasi addresses this gap directly. When a genomic panel result is uploaded or pulled via FHIR from a laboratory partner, our platform automatically screens for actionable variants and surfaced a structured alert within 87 seconds of data receipt. The alert does not require the oncologist to read the full report. It says: "This patient has an ERBB2 amplification. This is actionable in the current diagnosis. Relevant FDA-approved therapies include [list]. Consider tumor board review." The goal is to make the path from genomic data to clinical decision as short as possible. As we discuss in our article on reducing diagnostic delays, time between data receipt and clinical action is where lives are most directly at stake.