Skip to main content


Pancreatic Cancer Progress? Maybe

There’s been a vast amount of work (and excitement) in the immunotherapy approaches to cancer in the last few years. Here’s a report on one that I’d missed: IMM-101, which is not directed to PD-1 or PD-1L1, not an engineered T-cell, or any of the other things that have been in the news. It’s a killed preparation of the bacterium Mycobacterium obuense, which mixture seems to have broad immunomodulatory activity.

This paper details the clinical results of this agent added to standard-of-care for pancreatic ductal adenocarcinoma. That, as many will recognize is one of those cancers with hideous one-year survival rates – very bad news indeed. And there’s not a lot that slows it down. But this trial in 75 patients (35 controls got gemcitabine alone) showed an interesting result. You might not think so, to look at the bottom line, which was an overall non-significant survival benefit. But the predefined metastatic group actually did show what could be a significant effect, and that’s unfortunately cancelled out by the localized tumor patients, who actually died more quickly in the treatment group. There have been some headlines about this study, but (perhaps understandably) none of them mention that last part.

A similar effect (greater efficacy against metastases) had been seen in preclinical studies on mice and in cell culture, which is surely why this Phase II trial had the predefined patient groups. But I don’t think anyone expected the reverse effect in the nonmetastatic patients, if indeed that’s real. Informed by this, further trials will look at metastatic patients alone. It’s worth noting that the gemcitabine control group had survival numbers that were a little on the low side, compared to what was expected, and that gemcitabine alone is not (by now) necessarily what you’d compare against (the patients in this study started enrolling in 2011). But this does look worth a follow-up, especially since IMM-101 seems to have added no particular toxicity or adverse events in the treatment group. Hold the headlines for a bit, though. . .

25 comments on “Pancreatic Cancer Progress? Maybe”

  1. Woese is me says:

    My bacteriology has weakened. Is there anything scientifically that differentiates this from Freund’s Adjuvant, beyond patentability?

    1. Derek Lowe says:

      Well, Freund’s has fairly nasty effects and can’t be used in humans, IIRC, and this one made it through Phase I just fine, so there is that.

      1. woese is me says:

        I suspect if you formulated this in research grade-mineral oil to provide a long-term depot, one would see similar toxicity. Regardless, this is probably a more compelling approach than using intravenous L monocytogenes in Pancreatic Cancer, though i’m not putting my IRA into either one.

    2. tangent says:

      BCG M. bovis is a major treatment for bladder cancer, has been for thirty years and I’m sure it’s been tried everywhere else with no breakthrough. Tantalizing.

      I’m curious, what is it about the mycobacterial immunological profile that makes them specially interesting as an adjuvant? What is the response they elicit?

      1. Woese is me says:

        I don’t know specific on why Myco and not other species. My no-so insigful guess is that I attribute it to the immune system reacting to components of ‘common’ pathogen, via TLR responses, etc. Bioniche has been trying to develop a safer BCG replacement for bladder cancer that is a more defined composition of just Myco cell wall components (i believe). If i recall they have has some efficacy problems, but i’d guess that they may have to be superior to BCG. i’m not sure where else BCG has been used, since it’s delivery is a bit challenging. (e.g., i’d hesitate to inject it into someone’s brain, even if they had a tumor; bladder instillation seems much more reasonable). The idea of using a non-specific immune stimulator applied intradermally, containing no tumor specific information, introduced away from the site of the tumor is not a convincing mechanism to me. Using BCG intravesically or other ‘intratumoral’ immune stimulants is moreso especially in light of other immune clinical data in the last few years.

  2. Anon says:

    So one group died slower, and the other faster? Let’s consider this more closely:

    Probability that both groups would appear to react exactly the same? Close to 0%.

    Thus, probability that one group would appear to react better than the other (whether the observed effect is real or not)? Close to 100%

    This is not an interesting breakthrough; it’s the inevitable outcome of random chance due to multiple hypothesis testing.

  3. idiotraptor says:

    Re: Anon’s propositions and conclusion above.

    I don’t know anything about the design, patient enrollment or treatments of the clinical trial cited above or how it was statistically powered. In the absence of presenting and examining these parameters, isn’t it a wee bit dismissive to breezily assert the results are “the inevitable outcome of random chance due to multiple hypothesis testing”?

    1. Anon says:

      Happy to make a bet if you are.

      1. Phil says:

        Whether Anon is right or wrong is beside the point. idiotraptor is pointing out that you are playing fast and loose with your stats and do not have enough information to support your assertion.

        You could very well be right, but it would be because you’re lucky, not good.

        1. Anon says:

          See my reply to your comment below. It’s very borderline once you take into account multiple hypothesis testing.

    2. Phil says:

      Chrispy linked the study below (thanks!).

      They only report two tested hypotheses (whether there were other subgroups they looked at I cannot say). The two p-values are 0.07 and 0.01. They claim the subgroup that saw a significant (p<0.05) response was pre-defined (metastatic). I don't see any evidence of cherry-picking subgroups to dig up a significant response.

      1. Anon says:

        If you test 2 subgroups (plus the combined group) for significance at p<0.05, then you actually need to test each group and subgroup for significance at p<0.05/(2+1)=0.017 (see Bonferroni correction), so this result at p=0.01 just barely qualifies as significant, but that is cutting it really fine. And that's assuming they didn't test any other subgroups but didn't report on them. Thus overall I would say it's just a result of pot luck due to random sampling.

        1. PS. says:

          … especially if you take into account publication bias.

        2. Phil says:

          @Anon: They reported just one subgroup (p=0.01) and the combined group (p=0.07). Regardless, your reasoning given the data implies that it’s a tossup, whereas your original statement reflected near certainty of no real correlation.

          I stick with my original statement, you could be right. But you don’t have enough evidence to be so confident.

          FWIW, I am sympathetic to your skepticism.

          1. Anon says:

            Has anyone ever seen one of these early tossup cases (p-value close to significance threshold) that *did* turn out to be statistically significant and reproducible in bigger and later trials? For some reason, what we would expect to have a 50% chance of success, turns out positive much less than 50% of the time. Clearly we are missing something…

  4. Mol Biologist says:

    I cannot ignore that the overall survival benefit for the 110 patients was impressive. Those who had IMM-101 survived a median of 6.7 months compared with 5.6 months for those just on chemotherapy.
    BACK to the ROOTS. Cancer is metabolic disease where cells are obtaining energy by unusual way. It affects whole body by dis-regulation from one metabolites primarily glucose and glutamine, to other metabolites, primarily ketone bodies.
    Whole body become indebted and many patients feel tired similar to sleeping sickness disease patients.
    Immune system needs a lot of energy to be activated in full capacity where mostly cancer patients suffered a deficit. The immunotherapy approaches cannot work efficient when whole system indebted because first you have to restore imbalance.
    Graig Tompson at least raises his voice and at least offers new treatments options to use drugs not previously thought of in the context of cancer — such as the diabetes drug metformin —to fight against the disease.

  5. Chrispy says:

    Did you mean to link to the paper? I assume it is this one:

    All the breathless headlines make the actual paper hard to find by Google!

  6. Not A Chemist says:

    “Those who had IMM-101 survived a median of 6.7 months compared with 5.6 months for those just on chemotherapy.”

    Is dying of cancer in 5.6 months better or worse than dying of cancer in 6.7 months? I’m guessing the distribution of answers will be Europe (no) versus the US (yes).

  7. MIke B. says:

    Sorry, I’m not paying $150,000 to live an extra six months. Just. Not. Worth. It.

    1. HT says:

      It’s a difference of 1.1 mth, more like 5 wks.

      1. steve says:

        Initial cancer trials are usually done on patients who have failed all other treatments. These treatments (chemo, radiation, etc) select for resistant subpopulations and so are very difficult to treat. It’s therefore very unusual for a novel treatment to do more than add a few weeks or months to the lifespan of those patients and almost unheard of in pancreatic cancer, one of the toughest to treat. In this case, the patients all had inoperable pancreatic cancer that had locally advanced (probably to the liver) or metastatic. Any treatment that does increase lifespan in patients with advanced cancer, may be much more effective early on or in combination with other drugs (this trial looked at the combination with gemcitabine, which has marginal activity at best). Most advances in the treatment of cancer has come from finding synergistic combinations of drugs that individually only added incrementally to life expectancy. This only comes from years of trial and error after a drug is approved.

        1. Barry says:

          the great advance in small-molecule oncology will probably be regulatory. Today, we test new potential onco drugs in patients who have already failed all approved therapies, whose cancers are known to be resistant and who have many, many driving mutations.
          We still haven’t taken to heart the lessons of Gleevec. It was tested in CML, for which there was no approved treatment, therefore could be tested in early-stage patients with only one mutation and it was wonderfully efficacious. In later-stage CML (“blast crisis”) where there are many more mutations driving, vastly less efficacious.
          To generalize this lesson, we need to be defining conditions with less than all of the “Hallmarks of Cancer” as disease (may need a new name for it). It is in these that our small-molecule targeted therapeutics are going to be most efficacious. (including some that have already crapped out in Clinical trials against advanced recalcitrant populations)

          1. wlm says:

            Except for skin and colon cancer, where we can generally remove pre-cancerous growths, what cancers give us the early diagnosis this approach would require?

  8. Mol Biologist says:

    IMO the efficacy may come from double efforts. First, the small molecule application to stop growing tumor energy demands. Second is more complicated since you need to restore whole body energy demand which came negative by paying tumor obligations. And here even more interesting. As each person is a unique metabolic entity, personalization of metabolic therapy as a cancer treatment strategy will require fine-tuning based on an understanding of human genomics.

  9. KeithD says:

    In 2009, NCI recommended patients with metastatic or locally advanced pancreatic cancer should be studied separately. Biology > stats?
    Philip et al: Consensus report of the national cancer institute clinical trials planning meeting on pancreas cancer treatment. J Clin Oncol 27:5660-5669, 2009

Comments are closed.