By sequencing more than 1,000 lung cancers, the Clinical Lung Cancer Genome Project, a large international consortium led by the University of Cologne in Germany, has made a comprehensive molecular classification of lung cancers which maps closely to some histological or pathological subtypes of the disease, but also suggests that others are false amalgamations of different distinct genomic groups.
The group's study, published last week in Science Translational Medicine, provided evidence from a prospective analysis of more than 5,000 patients that using genomic information about a patient's tumor to guide treatment leads to improvement in overall survival for the subset of tumors with actionable alterations like EGFR mutations or ALK rearrangements.
According to the study's first author, Roman Thomas, the results indicate the potential benefits of a broad implementation of genome-based diagnosis strategies in lung cancer.
"We showed that outside a defined clinical trial, just genotyping a very big group of solid tumor patients and trying to address specific genomic alterations improves survival," Reinhard Büttner, a co-author on the study and member of the University of Cologne team, added.
While the study did uphold the accuracy of histology and morphology in defining some lung cancer subsets, Thomas said, it also clearly pointed out the added value of genomic information. "An adenocarcinoma per-se does not constitute a therapeutically relevant subgroup. However, if you know that it is a KRAS mutant, now you could potentially enroll this patient in a trial with a new compound targeting KRAS. The genomics really make a difference there," he said.
Büttner told Clinical Sequencing News that the international project had two main sections. In the first, the researchers retrospectively analyzed a large set of clinically-annotated lung tumors across all histological subgroups using exome and targeted sequencing to identify genetically-defined subtypes and measure how well those corresponded to traditional histological categories.
For some categories, like adenocarcinoma and squamous cell carcinoma, the team was able to identify patterns of genomic alterations that matched closely with histological groups.
"Traditionally, you have non-small cell lung cancer and small cell lung cancer, and within non-small cell, you have adenocarcinoma, squamous cell carcinoma, and large cell carcinoma," Thomas said.
"From our results, it turned out that adenocarcinoma is a robust subtype in whatever space you look, whether mutations, amplification, gene-expression, or pathology. And the same was true for squamous cell carcinomas," he said. "Under the microscope, it can be distinguished, but [these cancers] have also distinct patterns of mutations, gene expression, and so on."
Importantly, he said, genomic information offers the opportunity to further subdivide these larger subsets into smaller, more specific groups, like KRAS-mutated or KRAS wild-type adenocarcinomas, for example.
Meanwhile, the sequencing results also showed that some other histological categories do not conform to distinct molecular groups. Large cell cancers were almost completely eliminated as a distinct subgroup in the team's genomic reclassification.
According to the results, the large cell category actually makes up a catch-all, including mostly tumors that are members of one of several of the other genomic subtypes. Less than two percent of what would be considered large cell cancers based on morphology were unclassifiable into other molecular subtypes, according to the CLCGP researchers.
"Pathologists have known this for a few years now," Thomas said. "Using IHC, people were seeing protein expression markers in large cell carcinomas that were compatible with other subtypes, and now we found that this is perfectly matched with the genetic data."
"And for large cell tumors that could not be differentiated using IHC, most of those could also be assigned to any of the other subtypes using genetics," Thomas added.
In the second part of the project, the team put their strategy of looking at both molecular and histological classifications into practice, using targeted sequencing to genotype patients for a variety of different clinically-relevant alterations and measuring whether that approach led to better treatment outcomes.
According to Thomas, the Cologne group has been using an amplicon-based Illumina sequencing strategy coupled with FISH for this clinical sequencing effort, but starting in the first quarter next year, the team plans to move to a broader capture-based, pan-cancer, targeted sequencing panel covering about 200 genes, which will allow them to deal with amplifications and fusions without the need for FISH.
"Some people are saying 'why not do whole exome?' But the problem is we cannot do this with the coverage we need. Lung tumors are very heterogeneous and you see some mutations only if you go for several thousand-fold coverage," Büttner added. "No one can do a whole exome at 2,000-fold in a reasonable period of time. In lung cancer you need to start therapies right away. You don’t have half a year to run an exploratory sequencing program."
In the study, the researchers used their genomic-based diagnostic approach in 5,145 lung cancer patients recruited over a three-year period through an outreach program in the region of the University of Cologne cancer canter — of which 75 percent, or 3,863 were able to be sequenced, yielding 1,481 genomic alterations.
Overall, the group found that patients who were genotyped using this strategy showed stage- and histology-independent improvements in survival compared to those whose tumors could not be analyzed due to complications like insufficient tissue. And, for the small subset of patients who were diagnosed as having EGFR mutations or ALK-rearranged cancers and who were treated with targeted therapies based on this diagnosis, the approach led to improved overall survival.
Sixty-four of 84 patients with confirmed EGFR mutations received treatment with erlotinib or gefitinib, according to the study authors, and 15 of 30 with ALK translocations received crizotinib, the group reported. Another 34 patients with BRAF, KRAS, or FGFR1 alterations were enrolled in clinical trials, the authors wrote.
According to Büttner, the results were strong enough that the Cologne team believes that in-depth genotyping should be performed as a routine diagnostic step in every lung cancer, and that where targetable alterations are identified patients should receive the appropriate drugs as a first-line treatment.
He said the team is now planning to expand its sequencing approach to melanomas and to gastrointestinal cancers in studies similar to the recently published lung cancer analysis.
"After we screen maybe a thousand patients we hope we can show that this concept is a blueprint for all of the solid cancers," he said.
During the course of the lung cancer classification project, Thomas and Büttner's team also discovered some promising drug targets specifically in small cell lung cancer, which they published in 2012 (CSN 9/5/2012)
Based on those results the group said at the time that it was developing a clinical trial treating small cell lung cancer patients with an experimental FGFR1 inhibitor. Büttner said that this phase I trial — which is evaluating a Novartis drug — recently finished recruiting and should be completed by the end of the year.