We thank Stricker and Johnson for their comments on our cohort study based on a linked Swedish registry data on the association between anti-Borrelia IgG or IgM seropositivity and the risk of malignancy. As pointed out in their letter, similar to most other tests, the ELISA test used in our study (IDEIA Borrelia burgdorferi IgM/IgG Oxoid, Cambridgeshire, UK) has limitations when it comes to sensitivity. In fact, prior studies have estimated the sensitivity for this method to be 64–69% for IgM and 50–77% for IgG.1, 2 Importantly, the recent Scandinavian case-control study, which observed a statistically significant association between anti-Borrelia seropositivity and risk of mantle cell lymphoma used a similar in-house ELISA test3 with the same estimated sensitivity. On a statistical note, low sensitivity of a test generally leads to nondifferential misclassification leading to bias toward the null value.4 Thus, it is plausible that we would have missed weak associations with Borrelia. What is reassuring is that the specificities are relatively high for this test at 92–95% for IgM and 98–100% for IgG,1, 2 which minimizes further nondifferential misclassification.
With regard to the inverse associations with lung and cervical cancers and positive association with prostate cancer, we were not convinced that these associations were real. First, as stated in our original manuscript, the lack of statistically significant associations after adjustment for multiple comparisons (adjusted p-values > 0.05) suggests that these observations are spurious. Additionally, as previously stated, we found no dose-response when estimating risk of malignancy at increasing antibody tertiles. Because we do not consider these true associations, we would not go further to consider a heterogeneous effect of Borrelia on the risk of malignancy.
As with any study, our study has limitations and the results need to be validated in future investigations. At this time, to our knowledge, our study is the largest systematic evaluation of anti-Borrelia seropositivity and the risk of developing solid or hematologic cancers. An important aspect to keep in mind when interpreting our results is that our testing methods are virtually the same as the Scandinavian case-control study which reported a statistically significant association between anti-Borrelia seropositivity and the risk of developing mantle cell lymphoma.3 Added advantages in our study include the prospective study design, which eliminates recall bias, and the ability to evaluate the risk of all malignancies. We carried out a thorough analysis and interpreted our results with caution. Taken together, we believe that our study represents an important contribution to the literature on Borrelia and malignancy.