Dr. Sonia Dacic and the lung pathology group at University of Pittsburgh Medical Center recently published in Modern Pathology an article (full text here), "Clinicopathologic predictors of EGFR/KRAS mutational status in primary lung adenocarcinomas." The authors state that the aim of the study was "to determine whether clinicopathological characteristics and morphology of lung adenocarcinomas might be used as predictors of tumor mutational status, which then may be implemented as the selection criteria for molecular profiling of lung adenocarcinomas in clinical practice." Implicit in this statement are the assumptions that molecular testing of lung adenocarcinoma is expensive and that screening cases for molecular testing by morphological criteria would be cost-effective.
This is a single institution study of 345 previously untreated, surgically resected lung adenocarcinoma cases from UPMC. One important methodological issue is the uniform histological assessment of every tumor using 2004 WHO classification criteria and assessment of percentages of various histological subtypes of adenocarcinomas. All tumors underwent KRAS and EGFR mutational analysis and assessment of EGFR amplification by FISH. One potential weakness of this study with respect to its possible general application is the marked over-representation of stage I tumors (201 out of 345) and slight over-representation of women (57%) and never smokers (12%) compared to the general population--indicating a possible referral bias. All three groups would be expected to enrich the study population for EGFR mutations/amplification. However, the frequencies of EGFR and KRAS mutations are similar to other studies (about 10% EGFR mutated and 30% KRAS mutated tumors). One weakness of this study (which the authors point out as well) is the lack of survival analysis--however, given the overabundance of stage I tumors, I'm not sure how generally applicable that data would be unless they focused on finding stage I subgroups with different survival outcomes. Their data also confirmed previously reported associations between EGFR mutations and female sex and never smoker history.
- The absence of solid growth pattern was a significant predictor of EGFR mutation, while mucinous growth pattern was a significant predictor of KRAS mutation.
- A prominent lymphocytic host response was significantly associated with KRAS+ and EGFR-/KRAS- tumors than EGFR+ tumors.
- There was no correlation between morphology and EGFR FISH positivity.
The authors' findings are consistent with previous studies of U.S. patients but the association found between lymphocytic host response and EGFR mutation status is a novel and interesting finding. The issues of TILs and peritumoral lymphocytic host response are controversial "hot topics" in solid tumor research and warrants further investigation in lung cancer research.
There is wide variation in testing for mutations in lung cancer ranging from institutions that test all lung adenocarcinomas to those which test upon request. I agree with the authors about the need for an algorithmic approach to testing, although I disagree about a "universal" one. Currently, one can propose several different algorithms that would each be "rational" and "evidence-based" but I don't think one would prove superior to the other, especially when deployed in a wide variety of settings. With mutation-specific MoAbs coming online for IHC, this picture should be even more textured. Back to the paper, none of the clinicopathologic criteria can be used to select lung adenocarcinoma cases for targeted testing due to the morphologic heterogeniety of lung adenocarcinoma itself and the large overlap between tumors with and without EGFR and KRAS mutations (just the two most common mutations!).