Dr. Weeks’ Comment:  the merits of Metformin and a low sugar diet continue to impress.

ATLANTA — Previous findings that diabetics have longer survival after surgery for non-small cell lung cancer (NSCLC) probably reflect use of the drug metformin, a researcher said here.

In a multivariate analysis of more than 400 patients with NSCLC, the odds of survival were three times greater in metformin users compared with nonusers, whereas there was no difference in survival between diabetics and nondiabetics, according to Samjot Singh Dhillon, MD, of Roswell Park Cancer Institute in Buffalo, N.Y.

He told attendees at the American College of Chest Physicians annual meeting that the findings “support consideration of metformin use in a prospective study” as an adjuvant treatment for lung cancer.

Overall, the literature on diabetes and lung cancer has been equivocal, Dhillon said. Diabetes appears to be a clear risk factor for development of lung cancer, with one study showing that some 26% of people diagnosed with stage I or II NSCLC also have diabetes.

But the effect of diabetes on survival after diagnosis has been harder to pin down. Among seven studies reviewed by Dhillon and colleagues, two showed a definite reduction in mortality, one showed an increase in risk of death, another found an increase in recurrence rates, and three yielded no difference in outcomes (although one of those found metastasis was less common in diabetics).

Metformin, meanwhile, has attracted significant attention as a possible antiproliferative agent. A recent meta-analysis showed dramatically decreased rates of both incidence and mortality for a group of cancers taken together.

But, Dhillon pointed out, only three of the studies included in the meta-analysis had data on lung cancer, and only on incidence, rather than survival after surgery.

To get a better handle on the question, Dhillon and colleagues analyzed 10 years of NSCLC patient records at Roswell Park. Metformin use was identified from pharmacy records, and patients’ charts indicated whether they had diabetes, as well as giving other information.

The analysis was restricted to patients with pathologic stage I disease at diagnosis who underwent lobectomy or more extensive surgical therapy. Patients with multiple primary tumors, neoadjuvant chemotherapy, or resections more localized than lobectomy were excluded.

Of nearly 3,400 patients with NSCLC treated at Roswell Park from 2002 to 2011, 409 met all criteria for inclusion. They included 71 with diabetes and 41 using metformin at some point. At the most recent follow-up, 257 of the patients were still alive.

There were few significant differences in patient and disease characteristics between metformin users and nonusers, or even differences with strong trends in one direction or the other. Users and nonusers were very similar in age, gender, race, smoking history, type of surgery, tumor histology, and pathological stage.

Insulin use, not surprisingly, differed significantly — only 2% of metformin nonusers were taking insulin, compared with 24% of users.

The only other major difference was the foundation for the study’s top-line result: 83% of metformin users were alive at follow-up, versus 61% of nonusers (P=0.005).

Dhillon said the odds ratio for death in nonusers versus users was 3.08 in multivariate analysis (95% CI 1.32 to 7.19, P=0.009).

Other factors associated with improved survival in the analysis included no history of smoking and an adenocarcinoma histology (versus squamous cell).

The researchers also examined whether the improved survival with metformin use might be explained by increased mortality in diabetics not treated with the drug. That appeared not to be the case, Dhillon said.

Survival in the 35 such patients in the cohort differed little from survival in the 333 patients without diabetes and not taking metformin (55 versus 61 months, respectively, P=0.147).