To the Editor:

Berger et al¹ report on the relationship between postresection CA 19-9 levels and survival in patients with pancreatic cancer treated with adjuvant chemoradiotherapy in the Radiation Therapy Oncology Group 9704 trial. In this analysis, the authors assigned a CA 19-9 level of 0 to patients whose red cell phenotyping for both Lewis A and B antigens was reported to be negative. They based this decision on a 1987 study by Tempero et al,² which they interpreted to show that “by definition, [such patients] do not have the ability to secrete CA 19-9 into their serum”¹; However, Tempero et al examined, in all, twenty patients, only three of whom were Lewis antigen A and B negative. The fact that none of these three patients had elevated CA 19-9 levels (they were not 0, but 4, 10, and 14) does not prove that all patients who lack Lewis antigens lack the ability to secrete CA 19-9. Referring to a table of binomial distributions tells us that the 95% CI for the proportion of patients with elevated CA 19-9 when none are found in a sample of three is 0% to 71%.³ Our usual statistical guidelines do not even permit us to suggest that Lewis antigen–negative patients are less likely to have high CA 19-9 levels than to have low levels.

The zeroing out of Lewis antigen–negative patients is all the more disturbing given the discordance Berger et al¹ note between the 34% incidence of such patients in their study and the 5% to 10% incidence in the general population.

Berger et al¹ appear to have missed an opportunity to further our understanding of the correlation between Lewis antigen blood group and CA 19-9 by failing to replicate the work of Tempero et al² with adequate quality controls. As a result, they may have confounded the results of their study of the relationship between postresection CA 19-9 levels and survival in pancreatic carcinoma.

AUTHOR'S DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST

The author(s) indicated no potential conflicts of interest.