New molecular test predicts lung cancer survival better than staging

Non-small cell lung cancer (NSCLC) unfortunately recurs even more often than other cancers, even when early stage and apparently completely resected. The explanation, of course, is it wasn't completely resected: there was residual metastatic disease hiding out somewhere, but in such microscopic quantities as to be undetectable by our improving but still crude techniques (i.e., mediastinal lymph node survey and PET/CT scans).

For this reason, "staging" is itself an arbitrary and flawed system to guide treatment decisions. Example: many people with resected stage IA lung cancer are truly cured, but others will die of metastatic disease within a few years. But because more people with stage IA will be harmed by chemotherapy than helped overall, and the few who would be helped can't be identified, chemotherapy is provided to none, and those who would have benefited have their lives cut short instead. In oncology, that's just the way it is, but the possibility of avoiding such inadvertent mismanagement is the reason for the hype and hope around advances collectively called the "personalized" or "genomic" approach to cancer diagnosis and treatment.

Score one more for the revolution, with Johannes Kratz, Jianxing He, David Jablons et al's report in Lancet in March 2012 of the development of a molecular genetic test that, when applied to resected non-squamous non-small cell lung cancer (NSCLC), predicts 5-year survival better than the traditional staging system. More important, the assay could possibly lead to better individualized treatment decisions for early stage IA-B NSCLC, leading oncologists to provide chemotherapy to the patients with early-stage NSCLC who might benefit most from it.

What They Did

Researchers at U. of California-San Francisco (UCSF) developed a 14-gene assay that uses quantitative PCR analysis of resected non-squamous NSCLC (from 361 patients) to predict tumor recurrence (i.e., the likelihood of microscopic metastasis already having occurred at the time of resection). By correlating / logistically regressing various genetic mutations against the observed survival in patients, a model was created to predict survival based on the presence/absence of each mutations and their various combinations. (For the curious, the genes were BAG1, BRCA1, CDC6, CDK2AP1, ERBB3, FUT3, IL11, LCK, RND3, SH3BGR, and WNT3A).

The assay was then validated in 433 patients with stage I non-squamous NSCLC from Kaiser Permanente, and in 1006 Chinese patients with stage I-III disease that was resected and banked in the China Clinical Trials Consortium. All the patients were followed for five years. The studied variable was absolute time to death after resection (no soft "progression-free survival" here).

Pinpoint Genomics, which is in a financial relationship with the researchers at UCSF whose ultimate goal is likely to be marketing the assay or selling it to someone else for big bucks, funded the study in part (by providing materials, equipment, etc).

What They Found

The genetic assay predicted survival with more nuance and stratified better than simple staging. For patients in the Kaiser validation cohort (who all had stage I disease), 5-year survival was:

• 71% in the low-risk cohort (collection of lowest-risk mutations)

• 58% in the moderate-risk cohort

• 49% in the high-risk cohort

More impressively, the assay predicted 5-year survival better than staging among the Chinese cohort, many of whom (~40%) had resected stage II or III disease:

• 74% in the low-risk cohort

• 57% in the intermediate-risk cohort

• 45% in the high-risk cohort

This is a complex result to analyze due to the lumping together of patients with stage IA (24%) who wouldn't have gotten any chemo, with the rest who generally would have, but it's remarkable regardless. Within each stage (I, II, III) in the Chinese validation group, the assay stratified patients by survival likelihood robustly, with separation in the three Kaplan-Meier curves (in other words, it worked).

What It Means

Randomized trials are equivocal as to the benefit of chemotherapy in stage I non-small cell lung cancer (NSCLC). Chemotherapy is not given in stage IA disease, but is given in stage IB, although any benefit of that practice is also questioned. Further, as mentioned above, there are patients in all stages of NSCLC who will benefit from chemotherapy, and others who will not.

There is no way to identify who will benefit from chemotherapy, today. The National Comprehensive Cancer Network (NCCN) advises chemotherapy for stage IB patients with worrisome factors like tumor > 4 cm, minimal margins, poorly differentiated histology, and many other criteria, but these are all anecdotal, untested, unvalidated variables. Authors here (with a likely large financial stake in convincing us of this) make the convincing argument that their assay performed better objectively than such criteria, and therefore should be purchased and used immediately. Meanwhile, they're planning to test it prospectively in a randomized trial that will use the assay to determine whether patients with resected stage I disease get chemotherapy or not. Their assay looks promising--potentially even life-saving and practice-changing--and I wish them well, both in their careers and their likely future well-funded retirements to Tahiti.

Kratz JR et al. A practical molecular assay to predict survival in resected non-squamous, non-small-cell lung cancer: development and international validation studies. Lancet 2012;379:823-832.