We thank Dr. Vijfvinkel and colleagues for their interest in our brief report and appreciate the opportunity to address their comments.
Two previous studies demonstrated an association between low 2D:4D ratio (as a possible marker for prenatal testosterone) and knee OA (1, 2). This observed association could possibly be due to a direct effect of testosterone on the cartilage in the prenatal period (3). Previous studies have also shown that a low 2D:4D ratio is associated with more aggressive and risk-taking behavior and with better athletic skills (4, 5), and the association with knee OA could therefore be mediated through a higher risk of knee injuries (which is a well-known risk factor for knee OA). We found an association between 2D:4D and previous self-reported knee injuries in men, but no association with meniscal damage or knee OA. This may be due to low frequency of posttraumatic OA and high frequency of degenerative meniscal lesions (not related to acute injuries) in middle-aged and elderly persons. Our goal was to gain a better understanding of risk factors for OA, and was primarily of academic character. Hence, we see no direct prevention opportunities with respect specifically to the 2D:4D length ratio, because the 2D:4D length ratio is possibly only indirectly and weakly associated with knee OA (no significant association was found in this study).
We used logistic regression to examine the association of the 2D:4D length ratio as the independent variable with knee OA, hand OA, meniscal damage, and knee injury as the dependent variables. We apologize if the description of the statistical methodology could be misinterpreted.
The participants in the Framingham community cohort were randomly selected from the general population (ages >50 years). Of those who were contacted, 1,830 of 2,582 expressed initial interest. Of those, 1,039 were examined, and 1,020 were included in this study. Further information about the recruitment and exclusion criteria is detailed in Figure 1 in another article by our group (6). The primary purpose was to establish a cohort representative of the general population in order to investigate the epidemiology of OA.
A meta-analysis showed that the sex difference in the 2D:4D length ratio was largest in the right hand and is possibly a better indicator of prenatal androgenization (7). However, the reasons for this side difference are unclear. We divided the participants into tertiles according to their 2D:4D length ratio, and the risk of misclassification of participants with left hand measurements was considered low, due to the use of only 3 categories. Exclusion of the data on these participants yielded similar results.
There are no established fixed-value boundaries for the categories with low, equal, and high 2D:4D length ratio. Several studies have relied on a visual classification, which is more prone to misclassification due to deviations of the hand. We therefore divided the participants into tertiles (i.e., the same approach as in earlier studies) (1, 2). Mean values for the length ratio of the phalangeal and metacarpal bones in the 3 visual classification groups have previously been reported by Robertson et al (8). In order to categorize the participants according to these values, we used the average of the mean values in those with low and equal 2D:4D length ratios as the first threshold and the average of the mean values in those with equal and high 2D:4D length ratios as the second threshold. The agreement between results obtained using these categories and those obtained using the tertiles was good for both the phalangeal and metacarpal 2D:4D (κ = 0.72 and κ = 0.69, respectively). Analyses using these thresholds yielded similar results as were obtained in our primary model (as presented in the report), but the association between low 2D:4D length ratio and knee injury in men was attenuated (odds ratio 1.53 [95% confidence interval 0.80–2.91]).