Editor's Note: An earlier version of this article mistakenly referred to Merck KGaA as Merck -- they are two different companies. The Motley Fool regrets the error.

With the launch of its phase 3 START2 study of tecemotide in non-small cell lung cancer, Merck KGaA (not to be confused with the U.S.-based Merck) is either advancing a valuable asset, or throwing good money after bad, depending on whether you are a believer in a subgroup analysis of its previously failed START trial. In this case, there are some compelling reasons to be a believer.

Merck licensed the drug from Oncothyreon (NASDAQ: ONTY) and tested it in a phase 3 trial of patients with unresectable, locally advanced Stage III NSCLC that had not progressed after initial chemoradiotherapy (CRT). Two-thirds of patients who received tecemotide received concurrent CRT, and one third received sequential CRT. In sequential CRT, patients receive chemotherapy first, and radiation afterward.

That trial missed its primary endpoint of overall survival. Median overall survival was 25.6 months for the tecemotide group, compared to 22.3 months for the placebo group, with a p-value of 0.123, which means the difference of 3.3 months did not meet statistical significance.

However, a subgroup analysis of 806 patients who received concurrent CRT showed a healthy overall survival benefit of 30.8 months compared to 20.6 months for the placebo group with a p value of 0.016, showing statistical significance.

About that subgroup
Subgroup analyses are often viewed with suspicion by investors. Pulling a positive result from a subgroup analysis out of an overall trial failure is not at all the same as having a successful trial, and the validity of such analysis have been questioned strongly in the scientific literature. It is often viewed as a way of finding a fit in the data for a desired result that can not be reproduced in subsequent studies.

However, Oncothyreon's analysis has a couple of things going for it. First, the subgroup was predefined at the start of the trial. It always intended to divide patients into concurrent and sequential CRT groups and compare those groups. That is unlike the sort of questionable post hoc analysis where the researchers select patients after the trial is completed for certain attributes and then only report groups who turn out to have positive results.

Second, Oncothyreon's concurrent CRT subgroup is quite large, representing the majority of the patients tested. The size of the group lends additional validity to the analysis. The significance of concurrent CRT versus sequential CRT is not known, but the size of the benefit measured for the Oncothyreon drug is impressive at over ten months, or nearly a 50% gain in survival. That's the kind of survival benefit makers of cancer drugs drool over.

It is hypothesized that concurrent CRT may lead to a type of cell death response that is different than the immunosuppressive effects of chemo and radiation given separately, and therefore more amenable to an immune-mediating therapy like tecemotide. It is also possible that the split between concurrent and sequential CRT separated the patients who were doing better from those not expected to achieve a cure. Immunotherapy products are thought to work better in patients with healthier immune systems.

NSCLC market is robust and growing
Merck partnered with Oncothyreon in 2001 in a 15-year, $150 million global collaboration. The deal has since been restructured with Merck responsible for all development costs while Oncothyreon is eligible for milestone payments and royalties. Oncothyreon could collect up to an additional $90 million in milestone payments for tecemotid, as well as royalties on worldwide sales.

The NSCLC market is highly competitive, currently led by Avastin (Genentech/Roche) and Tarceva (Astellas Pharma). However, according to a report by GlobalData, new immunotherapy products are expected to eventually outsell those market leaders. It projects 2022 revenues of $688 million and $348 million, respectively, for Boehringer Ingelheim's pipeline drug afatinib and Eli Lilly's necitumumab, and $1.75 billion for Bristol-Myers Squibb's investigational immunotherapy for NSCLC, nivolumab.

The overall NSCLC market is expected to grow at 3.3 percent annually to $7.9 billion in 2022. That means Merck's investment in tecemotide could pay off big time in the long run, if it can get the results it wants from the START2 trial. Although followup trials based on post hoc subgroup analyses have a poor track record generally, the large, predefined group combined with the very large treatment effect differentiate START2 from the sort of sunk cost fallacy situation some companies fall into when they try to massage the data from a clearly failed trial.

Priming the immune system for cancer
Tecemotide is designed to stimulate the immune system to attack cells with a surface antigen called MUC1, which is commonly found in NSCLC and some other cancers. Its strategy is therefore similar to nivolumab's, which is designed to attack cells that express PD-1, another marker commonly found on NSCLC cells.

BMS plans to use nivolumab as a second line therapy, whereas tecemotide is developing as a maintenance treatment. It's not clear yet what effect a concurrent CRT prerequisite will have on the market and sales potential of the drug.

Merck is carrying out the START2 trial under a special protocol assessment with the FDA, which makes approval highly likely if the trial is successful. Merck's progress forward with tecemotide makes Oncothyreon a potential acquisition target, as it would gain back the remaining milestone payments and royalty stream due on the drug. Its follow-on cancer vaccine, ONT-10, also targeted to MUC-1 is an additional asset in Oncothyreon's pipeline that could have future value.

Although Merck's investment in a second phase 3 trial after the unequivocal failure of the first may look like madness, in this case there is clearly method in it, and a good chance of success for the drug.