Research Signal – Source date 2021-01-01
When cancer spreads, doctors may have a tissue sample from the original tumor but not from the metastasis. A metastatic lesion can be deep in the brain or spine, for example, or a person may be too unwell for a new procedure. The first sample may therefore be the only available guide to markers used in considering immunotherapy. But does it still describe the cancer that has spread?
A 2021 systematic review and meta-analysis compared immune-related markers in primary tumors and paired metastases. The results depended on which marker researchers measured. PD-L1, a protein measured to help guide immunotherapy, differed between paired sites more often than microsatellite-instability status did. Some immune-cell measurements also shifted substantially. These figures describe differences in test results; they do not show whether retesting changes treatment success.
PD-L1 could change in either direction
Across 38 studies involving 2,109 patients, the pooled rate of PD-L1 discordance was 22%. In other words, about one in five paired samples had different positive or negative classifications at the original and metastatic sites. The estimate’s 95% confidence interval was 18% to 26%.
The review also calculated the two directions separately. Among pairs where the primary tumor was PD-L1 positive, the pooled estimate for a negative result in the metastasis was 41%. Among pairs where the primary tumor was negative, the estimate for a positive result in the metastasis was 16%. Each percentage starts with a different group of pairs; they are not portions of all patients and should not be added together.
A positive-to-negative change means the original sample met a study’s criteria for a positive result while its matched metastasis did not. Negative-to-positive means the reverse. The review found positive-to-negative changes more often, but those pooled patterns do not establish what a particular person’s metastatic sample would show. The classification can also depend on how a study measured the marker and where it set the threshold for a positive result.
Immune cells in the tumor also varied
Tumor-infiltrating lymphocytes, or TILs, are immune cells found within tumor tissue. Across 12 studies and 333 patients, the pooled rate of a difference in TIL level between paired sites was 39%. Among pairs starting with a high TIL level in the primary tumor, the pooled high-to-low estimate was 50%; among those starting low, the low-to-high estimate was 16%.
The review also found pooled discordance estimates of 26% for PD-1 and 22% for PD-L2. These are additional markers in the immune checkpoint system, but the number of studies contributing data differed by marker: eight studies for PD-1 and four for PD-L2. Their estimates, like the PD-L1 and TIL figures, summarize the studies included rather than predict an individual result.
Genomic markers showed a different pattern
Microsatellite instability (MSI) is a genomic marker assessed in some tumors. Its status differed in a pooled 6% of paired samples across six studies involving 347 patients. The review’s directional estimates were 9% from MSI-high in the primary tumor to microsatellite stable (MSS) in the metastasis, and 0% in the reverse direction. Those estimates had wide uncertainty, particularly for MSI-high to MSS, and come from a smaller evidence set than the PD-L1 estimate.
Tumor mutational burden (TMB), another genomic measure, was summarized from five studies involving 75 patients. Most comparisons did not find a statistically significant difference between primary tumors and metastases. Some reported higher TMB in brain metastases, while others found no significant difference or a small difference in the other direction. With so few patients and studies, a lack of statistical significance cannot establish that TMB is identical across sites.
The contrast is useful, but it is not a guarantee that MSI or TMB will stay the same in every cancer. The studies measured different markers in different groups, and the available evidence for each one was uneven. A pooled rate describes the collection of paired samples in a study set, not a rule that applies to every tumor type or metastatic location.
How the pooled estimates were built
The authors screened 6,708 potential articles and included 56 studies with 2,739 patients in the systematic review. The studies compared markers in a primary solid tumor and a paired metastasis. For suitable categorical results, the authors pooled discordance rates using a random-effects model, which allows the underlying rates to vary across studies. TMB was summarized separately because the studies did not provide the same kind of categorical data for pooling.
The research covered nine cancer types, and studies varied in metastatic site, timing, assay method, scoring approach and positivity threshold. For example, PD-L1 can be assessed using different scoring methods, and a result near a study’s cutoff may be classified differently under another threshold. The review reported that dividing studies into subgroups by features such as assessment method and timing could reduce some of the variation in estimates. Few studies, however, reported enough data for detailed comparisons by specific metastatic site.
Most of the included studies were retrospective. Their paired samples capture a real difference in measured results, but the review could not isolate a single cause. Assay choices may contribute; time between samples, tumor changes and treatment history may also matter. The included reports did not consistently record treatment and paired-marker details in enough depth to determine how these factors affected conversion.
A changed result is not yet a changed outcome
A mismatch between two test results is not itself evidence that choosing treatment based on the metastatic sample would improve immunotherapy response or survival. The review measured marker discordance, not the effect of a testing strategy on patient outcomes. It also does not show that the pooled percentages apply to a specific patient, cancer type or metastatic site.
The authors recommended evaluating markers in both primary and metastatic tumors, while recognizing that metastatic tissue can be difficult to obtain. Whether retesting is feasible or appropriate depends on the clinical situation. The evidence available for this review did not establish a clinical benefit from retesting, and the paper noted that solid evidence or guidelines supporting reassessment of these immunotherapy markers were lacking at the time.
What testing a metastasis could change
If metastatic tissue can be obtained safely, and prospective studies show that its marker status meaningfully predicts checkpoint-treatment outcomes, clinicians could make treatment decisions using information that better represents the cancer being treated. The paired-sample findings make that possibility worth investigating, especially for markers that often differed between sites. They do not yet show that a second sample leads to better outcomes.
Key Sources
- Primary source: Discordance of immunotherapy response predictive biomarkers between primary lesions and paired metastases in tumours: A systematic review and meta-analysis.
DOI or Journal Page · Review or Meta-Analysis · 2021-01-01
Keep exploring