Red death and reference classes

I’d originally planned to write something this week on the announcement that the Nobel prize in Physiology/Medicine has been awarded to Campbell, ┼îmura and Tu. While there’s lots of possible interest here – the Neglected Tropical Disease angle, or the unusual military aspect to be found in the intellectual history of Tu’s work on artemisinin. However, I’ve been distracted by something that came out of S. Lochlann Jain’s excellent new-ish book Malignant: How Cancer Becomes Us, which I’ve been avidly reading this week.

The book deals with many themes but the take-home message that I’ve been most impressed by is the difficulty of navigating cancer. Jain argues that this is partly because cancer is so very complicated: it means different – possibly totally incompatible – things to different people. Cancer, she argues is a nexus of different experiences. That sounds about right to me, but rather than run through her argument for that here, I’ll try and pick out one part that I thought was especially interesting. This has not been an easy selection: her criticism of cancer conflict narratives – such as wars on cancer, or beating cancer – alone would make the book well worth reading. On these, you could also look up the excellent essay by Joanna Baines in the collection called Cancer Patients, Cancer Pathways: Historical and Social Perspectives which I reviewed last year in Studies in History and Philosophy of Science Part C. Alternatively, the discussion of some of the implications of randomisation (in chapter 5) are darkly fascinating, and packed full of historical nuggets of interest to the evidence enthusiast, like Alfredo Morabia’s article on a nearly-randomised trial on bloodletting that took place more than a hundred years before the usual ‘first’ example of randomisation in medicine: the 1948 MRC streptomycin trial for pulmonary tuberculosis (spoiler: it showed that bloodletting worked).

However, I was most struck by a (fairly brief) discussion of the drug epirubicin, which chimed with some of the work we’ve been doing on this project about reference classes. I’ve already talked a bit about reference classes on this blog, but epirubicin might be less familiar. This is one of the anthracycline drugs used for breast cancer chemotherapy. It has, Jain tells us, a chilling nickname in clinical practice, where it’s known by some as the red death. Red, because:

website imageDeath, because of the side effects. Even for cancer chemotherapy, where serious side-effects are considered almost normal, epirubicin has a fearsome reputation for causing harm. This reputation is so bad that the drug has developed something of a aura of an invincible weapon against some forms of cancer. As Jain writes of her own experiences

“Epirubicin became part of my self-image of toughing it out, toughing out the hardest possible chemotherapy because if I were tough enough, cancer would be scared away. I could handle the red death.” (Jain, 2014: 121)

Despite these side effects, evidence had been produced that appeared to show a survival benefit (or, more convincingly, the 2005 evidence overview) when epirubicin was given to women with stage II and III breast cancer. This benefit – while modest – meant that the drug became part of the standard regimen for some kinds of breast cancer in the late 1990s and early 2000s. Jain estimates that “nearly a million” people were exposed to it. However, it turns out that epirubicin is only of benefit in cases where the cancer displays certain genetic markers (HER2, especially). To quote an editorial from 2009 by Slamon and Press:

“Over the past 15 years, a substantial amount of clinical data from multiple individual studies has indicated that the incremental benefit from adjuvant anthracycline-based therapies is largely restricted to the HER2-positive subgroup of human breast cancers” Slamon and Press 2009

This is an excellent example of the way that important reference classes are often hard to detect in large trials. In this example, while epirubicin provided a modest average improvement in survival, this benefit was extremely unevenly distributed. Perhaps reference classes aren’t the most important part of the cancer nexus for Jain, but they chimes with something that we’ve been investigating as part of the EEiM project.