Introduction

It has been established that the index to ring finger ratio or 2D:4D ratio is a reliable biomarker of testosterone exposure during early intra-uterine life [1,2,3]. During the last decade, the 2D:4D ratio has been examined in relation to a number of physiological processes and psychological aspects [3], sporting abilities [4, 5] and diverse health conditions [6, 7].

Finger length ratio was also found to be associated with skeletal variables such as bone mineral density [8] and osteoarthritis (OA) in various sites of the body [9,10,11,12]. The type 3 finger length pattern of the index to ring finger ratio, i.e., low ratio or “male” pattern was found associated, to a statistically significant degree, with OA [10,11,12,13]. Sigurjonsdottir et al. [11] found that the type 3 finger ratio was associated with total knee replacements but not total hip replacements. The underlying mechanism for such findings is unknown.

We carried out a radiographic study on a large population sample who had participated in a Chuvashian skeletal aging study project, investigating different aspects of skeletal aging within a Chuvashian population [14]. The aim of this study was to evaluate the association between the index to ring finger (2D: 4D) length ratio and the severity of radiographic hand OA.

Methods

Study design

A cross-sectional population-based analytic study.

Sample

The population sampled was native Chuvashians residing in numerous small villages in the Chuvasha and Bashkortostan Autonomy regions of the Russian Federation. The Chuvashians participating in this investigation were randomly chosen volunteers described in detail elsewhere [15, 16]. The study cohort included 802 males, aged 18–89 years and 738 females, aged 18–90 years. The data collected included sex, age, anthropometrical characteristics (height, weight), occupation, as well as the nature and extent of their physical activities and details on their chronic morbidity and medical treatments. Excluded from the study were individuals with known bone diseases, amenorrhea, post-traumatic, rheumatoid or psoriatic arthritis, as well as intake of hormone replacement therapy or steroids. X-ray films of both hands were obtained from the study participants in addition to the necessary examinations, measurements, and interviews. All these procedures were consensual. The subjects signed an informed consent form. The entire project was approved by the Helsinki Ethics Committee, Tel Aviv University.

Body mass index (BMI)

BMI was computed as the ratio of weight (in kg) divided by height (in meters) squared.

Hand radiographs

Single plain radiographs of both hands using a standard roentgenographic technique as described in detail by Pavlovsky and Kobyliansky were taken in the posteroanterior position with an X-ray source located 60 cm above the patient [17, 18]. Hands were placed on the same film-containing plate to avoid any film or development variation and exposed for 5–10 s at 100–150 mA without intensifying screens at 50 kV. All X-rays were digitized. The radiographic measurements were performed using digital images.

Visual classification of finger length ratio

All X-rays were evaluated by an experienced radiological researcher (L.K.). Each hand was classified according to the length of the index finger; longer (Type 1), equal to (Type 2) or shorter than the ring finger (Type 3), by visual comparison of the soft tissue outline of the fingertips on the radiograph. As in a study of Robertson et al. [19], X-ray findings were classified as ‘definite’ or ‘probable’ according to the certainty of the observer. Each X-ray classified as “probable” was assessed by an additional reader and a consensus classification was recorded. Inter-reader correlation was high (κ = 0.86).

The data on actual finger (2D:4D) length ratio in Chuvash population was presented in details in our previous publication [20].

Radiographic assessment of OA

Each radiograph was read by an experienced and specially trained physician. The development of OA was evaluated for each of the 14 joints (4 DIP, 4 PIP, 5 MP, and IP-1) according to the Kellgren and Lawrence (K–L) grading scheme, which utilizes photographs from the Atlas of Standard Radiographs [21]. The extent of each of the above-mentioned indexes for each joint, ranged from 0 to 4. Joints scored as K–L ≥ 2 were considered affected. Each individual was characterized by the total number of affected joints and total OA score (total sum of K–L scores of all 28 joints of both hands). Summing up K–L scores of the hand joints in evaluating hand OA is common in epidemiological studies [22,23,24]. In addition, we divided all subjects to ones with vs. ones without hand OA (at least one joint with K–L ≥ 2).

Reliability of OA assessment

Initially, two experienced researchers (an orthopedic surgeon and a physician experienced in interpreting X-rays) read a batch of radiographs and decided on the protocol for evaluating the K–L scores. 10 X-rays were read using this protocol and then re-read by two separate investigators to estimate the intra- and inter-rater reliability of the readings. All discrepancies were reviewed for systematic errors. This exercise continued until the reliability was high (ICC > 0.8). Once high intra-rater and inter-rater reliability was established, one investigator still blinded to patient identifiers and patients’ ages, read all the X-rays. The intra-observer reliability (kappa statistics) was 0.85 (p < 0.01).

Statistical analysis

All statistical computations were performed using SPSS 17.0 for Windows (SPSS, Chicago, IL, USA). Mean and standard deviation for age and body mass index (BMI) measurements and frequencies of visual classification types were calculated separately, for each sex, using descriptive statistics. To compare the continuous variables (age, BMI, the number of affected joints and total OA score) between the males and females a one-way analysis of variance (ANOVA) was used; to compare the finger ratio types, the Pearson’s Chi-square test was used; to test the association between OA variables (number of affected joints and total OA score) and age and BMI, the Pearson correlation analysis was used; to test the association between OA variables (number of affected joints and total OA score) and age and BMI and to compare the OA variables between individuals with different finger length ratio types (after adjustment for age and BMI) a one-way analysis of covariance (ANCOVA) was used. These analyses were performed four times, separately for males and females and for finger length ratio types of right and left hands.

In addition, to test the association between finger length ratios and hand OA traits we performed two linear regression analyses, with number of affected joints and total OA score as dependent variables and age, sex, BMI and right and left finger ratios as independent predictors, and one bivariate logistic regression analysis with dichotomous variable of presence or absence of hand OA (at least one joint with K–L ≥ 2) as a dependent variable and same independent predictors.

Results

The study sample (Table 1) included 802 males (mean age 46.98 ± 17.10 years) and 783 females (mean age 48.65 ± 16.62 years) (age difference between males and females was not significant p = 0.053). The male BMI (23.19 ± 3.26 kg/m2) was lower than the female BMI (25.16 ± 4.87 kg/m2) (p < 0.001). Females exhibited a significantly higher number of affected joints (p = 0.039) and total OA score (p = 0.001).

Table 1 Demographic and clinical characteristics
Full size table

Distribution of finger length ratio types in the studied sample is shown in Table 2. No significant differences (p > 0.05) in the prevalence of each type of finger ratio in males and in females were shown between the right and left hand. However, when comparing the distribution of finger length ratio types between males and females, a significant difference was observed (d.f. = 2, χ 2 = 11.769, p = 0.003) in the right hand with no significant differences (d.f. = 2, χ 2 = 5.107, p = 0.078) in the left.

Table 2 Distribution of finger length ratio (2D:4D) in the studied sample
Full size table

Distribution of hand OA traits in the studied sample according to the age groups and sex is shown in Table 3. A number of affected joints, total OA score, and proportion of subjects with at least one affected joint, all gradually increased with age, especially after the 5th decade in males and females. Even, as we mentioned before, females had significantly higher number of affected joints than males (Table 1), in Table 3 one can see that until age 50, females had a lower number of affected joints than males, and the situation changed after 5th decade. Interestingly, almost all males after age 70 and females after age 60 had at least one joint with hand OA.

Table 3 Distribution of hand OA traits in the studied sample according to the age groups and sex
Full size table

Hand OA parameters showed significant positive correlations (p < 0.001 for both variables) with age and BMI (Table 4). Results of comparisons of OA variables between individuals with different finger length ratio types (after adjustment for age and BMI) are shown in Table 5. In unadjusted data, the highest values of a number of affected joints and the total OA score was seen in individuals (males and females) with a type 3 (2 < 4) finger ratio. However, when adjusted for age and BMI, significant differences were seen only in females between finger ratio types of the right hand in a number of affected joints (F = 3.153, p = 0.043) and finger ratio types of the left (F = 3.330, p = 0.036) and right (F = 2.397, p = 0.047) hands. Females with a type 3 ratio displayed the highest adjusted values of hand OA parameters.

Table 4 Results of the Pearson correlation analysis between hand OA parameters and age and BMI
Full size table
Table 5 Association between visual classifications of 2D:4D ratio and hand OA parameters
Full size table

Linear regression analyses showed that after adjustment for age, sex and BMI, right finger length ratio showed significant positive association with total OA score (beta = 0.38, p = 0.040) and with a number of affected joints (beta = 0.50, p = 0.046). No association was found between left finger length ratio and hand OA traits. In a logistic regression analysis, no association was found between the dichotomous variable of presence or absence of hand OA and finger length ratios.

Discussion

In this large population-based study, a significant association between type 3 finger ratios and the severity of radiographic hand OA in females was observed. The association is independent of other established risk factors for OA, such as age and BMI.

Previous studies have shown diverse results. A large population-based Icelandic study [11] showed no association between type 3 ratio and hand OA. In this study, hand OA was assessed by hand photographs, not X-rays, which may explain the different results. Haugen et al. [25] in a Framingham community cohort study, found a significant association between a low 2D:4D ratio (similar to a type 3 pattern) and hand OA in females, but not in males. Similarly, another large population-based study from the Netherlands [10] also found that the type 3 finger length pattern was associated with the prevalence of radiographic hand OA. Thus, results of our study support the previous findings of an association between hand OA and type 3 finger length ratio patterns in females. Furthermore, this is interesting since we used a different method for defining hand OA.

Theoretically, severe OA could lead to shortening of the digits and an overestimation of the presence of type 3 finger length patterns [10]. However, in our previous study performed on the same population [26], no differences were found in the degeneration rate or severity between the second and fourth fingers. de Kruijf et al. [10] found that when individuals with severe OA in the second finger were excluded, the association between hand OA and the type 3 finger length ratio, remained significant. The type 3 finger length pattern was also found to be associated with OA in the knees [10,11,12,13] and hips [13].

One of the possible explanations for the association between the type 3 finer ratio and OA is a hormonal mechanism [27]; estrogen, in particular, has a complex and possibly protective relationship with OA [28]. An association of type 3 finger length ratios with hand OA (not weight bearing joints) supports this theory [10]. Testosterone may also be involved in the etiology of OA. Ma et al. [29] conducted in vivo investigations on male mice (orchiectomy versus controls) and determined that intact mice demonstrated more severe OA. Both estrogen and testosterone are also known for their role in bone metabolism [30, 31], however, current evidence does not suggest that the finger length ratio is associated with hormonal levels in adults [32]. Further studies are warranted to understand the mechanism of the association between finger length ratio, especially type 3 and OA. In conclusion, the finger length ratio was found to be associated with hand OA parameters. Females with a type 3 finger length pattern showed significantly higher hand OA values than types 1 and 2.