Abraham Cherrix and the Fallacy of Moderation

Fellow ScienceBlogger Orac has posted the latest on Abraham Cherrix, the Virginia teenager who rejected conventional medical therapy for his relapsed Hodgkin lymphoma in favor of what we in the business call "spot-weld" radiation therapy, as well as unproven alternative supplements, presumably ingestibles. Since this story has piqued my interest I would like to comment on what appears to be a tragedy in the making.

Mr. Cherrix is back under the care of his radiation oncologist in Mississippi, getting XRT to a "small tumor on his right lung that showed up on a scan taken last week". He gives the following explanation for his relapse:

"When you consider that I've had cancer four or five times, and each time it's been an aggressive cancer with massive tumors or spreadout small tumors where there wasn't much we could do for treatment, when you have no cancer in your lymph system and a small tumor somewhere else, it's not really that big of a deal."

His mother is quoted as saying: "The cancer's not on the level that it was before, so what we are doing is working"; and "He's got a spot in his lung, but hey, look at the big picture here."

To me this line of reasoning is a good example of a logical fallacy called the Golden Mean Fallacy, or the Fallacy of Moderation.

This fallacy is committed when it is assumed that the middle position between two extremes must be correct simply because it is the middle position. This sort of "reasoning" has the following form:

Position A and B are two extreme positions.
C is a position that rests in the middle between A and B.
Therefore C is the correct position.

Using Mr. Cherrix's reasoning, then, position A is "The cancer has not returned, therefore my treatments have worked" and position B is "The cancer has relapsed throughout my body, therefore my treatments have failed." Position C, the middle ground position, would thus be "My cancer has only come back in one area, therefore my treatments have worked on most of my cancer." Why do I consider this to be fallacious logic?

The answer is because patients with active relapsed Hodgkin lymphoma do not live long; unlike other chronic illnesses Hodgkin disease kills if it is not killed itself. There is no middle ground; no partial credit is given for eliminating some of the tumor.

Now it is documented that patients who relapse after initial chemotherapy can be cured with salvage radiation therapy, but such treatment is only given to highly selected patients:

Patients enrolled in these studies are a highly selected patient population. For instance, in the study from Josting et al., only 100 patients (approximately 2%) eventually underwent this treatment approach out of 4,754 patients enrolled in the German Hodgkin Lymphoma Study Group trials HD4 -HD9. These studies suggest that patients who have early-stage disease with no extranodal sites of involvement, lack of B symptoms, and a good response to first-line therapy are the best candidates for this approach.

So long-term survival has been achieved with just radiation therapy alone. Could Mr. Cherrix be one of these lucky patients?

Of course he could, and I pray that he is one. The problem is that according to the above-mentioned excerpt he has at least two unfavorable prognostic features. The first is his new relapse in the lung, an extranodal site:

On multivariate analysis significant factors for FFS [failure-free survival] were B symptoms at the time of SRT [salvage radiation therapy] (p = 0.003), extranodal involvement (p = 0.011) and histology (p = 0.018)...Our data suggest that patients most suitable for SRT are those with relapse in supradiaphragmatic nodal sites and no B symptoms.

The other discouraging sign is that his initial duration of remission, from his diagnosis in the late summer of 2005 to his relapse in February 2006 is less than one year:

Radical RT [radiation therapy] is an effective salvage regimen for select patients with advanced stage Hodgkin's disease who relapse following initial treatment with chemotherapy alone provided that relapse is limited to sites which can be encompassed by radical RT fields and the DFI [disease-free interval] is greater than 12 months. Review of other published series supports DFI > 12 months as a favorable prognostic factor.

Even the results of chemotherapy (conventional or standard-dose, that is) after relapse from initial conventional chemotherapy are not great if the first complete remission was short:

If the time of relapse was within 1 year of obtaining a complete response, 49% achieved a second complete response. These remissions were not durable, and at 5 years, only 14% were disease-free. The median overall survival was 2.6 years for this group as a whole. This is in contrast to patients whose first complete remission lasted more than 12 months [my italics].

Thus, my final opinion: Abraham Cherrix should consider high-dose chemotherapy with stem cell support if he wants to cure his Hodgkin lymphoma. However, even this ultra-therapy has no guarantee of success, especially if one factors in his lung disease and his short duration of response to initial therapy:

Moskowitz et al. developed a prognostic model based on 65 consecutively treated patients with relapsed or refractory HD to predict the outcome from an autologous transplant. Three factors (i.e., extranodal sites of disease, complete response of less than 1-year duration or primary refractory disease, and B symptoms) predicted event-free survival. For patients with zero or one factor, the 5-year event-free and overall survivals were 83% and 90%, respectively. If two factors were present, the 5-year event-free and overall survivals were 27% and 57%, respectively, and this decreased to 10% and 25%, respectively, if all factors were present. In a study conducted by Stanford University, B symptoms, stage IV disease involving the bone marrow or lung, and greater-than-minimal residual disease at the time of presentation for transplantation were associated with a poor prognosis. The 3-year failure-from-progression was 85%, 57%, 41%, and <20% for zero, one, two, and three factors, respectively.

Sometimes it's hard to be cheerful in this line of work. Abraham Cherrix is a good example of why.