Response Free

We thank Engels and colleagues for their comments on our recent manuscript (1) and for highlighting their recent publication examining the association of a variety of medical conditions with the risk of non-Hodgkin lymphoma (NHL) (2). Although similar, our studies differ in several important respects that may have had bearing on the findings.

First, the control subjects in our study were matched to each chronic lymphocytic leukemia (CLL) case (up to five control subjects per CLL case) and were sampled without replacement whereas the control subjects in the analysis by Engels and colleagues could be selected more than once or included later as a case patient. Furthermore, in our multivariable analysis we also controlled for rurality, income quintile, and comorbid disease burden (Aggregated Diagnosis Groups) whereas Engels and colleagues did not. There may have also been differences in the CLL cases between the two studies as peripheral blood flow cytometry is not routinely reported to the Ontario Cancer Registry and, as such, our CLL population may have tended toward more advanced-stage disease. In the correspondence from Engels and colleagues, the NHL patients encompassed in their study included 19 078 diffuse large B-cell lymphoma (DLBCL), 18 236 CLL, and 8881 follicular lymphoma (FL) cases. We were surprised by the disproportionately high number of CLL cases given that the annualized age-standardized incidence of CLL is 4.6 per 100 000 (http://seer.cancer.gov/statfacts/html/clyl.html) vs 6.6 and 2.7 per 100 000 for DLBCL and FL, respectively (3). This raises the possibility that the patients with CLL in the analysis reported by Engels and colleagues may have also included other forms of lymphocytosis, circulating lymphomas, or leukemias—diluting out the true CLL cases for analysis. Finally, compared with our study, Engels and colleagues reported higher rates of dyslipidemia in their control subjects (49.2% vs 16.3%) and CLL populations (45.8% vs 22.2%), making it challenging to compare the two studies.

Importantly, the association of dyslipidemia with CLL has a biologic rationale. For example, there is increased expression of lipoprotein lipase (LPL) in CLL cells but not in normal lymphocytes, and a higher expression of LPL predicts for unmutated IGVH status and is associated with shorter treatment-free survival (4). In addition, lipid-activated nuclear receptors, particularly peroxisome proliferator activated receptor (PPAR)–alpha, a central regulator of fatty acid oxidation, and PPAR-delta, a regulator of mitochondrial efficiency, are overexpressed by CLL cells compared with other blood cancers, and PPAR-alpha and PPAR-delta antagonists kill CLL cells (5,6). Using an in vitro model, we recently demonstrated that low-density lipoproteins (LDLs) increased STAT3 phosphorylation. This in turn activated CLL cell numbers—an observation not seen in normal lymphocytes (7). The cholesterol content of circulating CLL cells also directly correlated with the blood LDL levels.

We designed our study to specifically examine the association of dyslipidemia and CLL and the role of antidyslipidemic agents on survival based on a biologic rationale. As a result, there were a number of methodological decisions made that differ from the analysis by Engels and colleagues, which was designed to screen for associations in a number of hematologic conditions. It is possible that any combination of these differences could have influenced these results, but we of course look forward to other studies independently investigating this issue.

References