Failed nutrient supply to the disc cells has been claimed to be the primary cause of intervertebral disc degeneration (DD)[1]. The metabolism of the avascular intervertebral disc is dependent on the diffusion of fluid which solutes into and out of the disc, and the disc cells do not survive prolonged exposure to low glucose concentration[2, 3]. Impaired nutrition could also explain the association between reduced lumbar artery blood flow and DD[4, 5].

Environmental factors such as smoking, obesity and physical inactivity may enhance development of DD by reducing blood flow. In experimental models nicotine caused stenosis of vascular buds, perivascular calcification, necrotic changes in endothelial cells and narrowing of the vascular lumen[6] and smoking induced DD[7], obesity increases the risk of atherosclerosis due to related atherogenic dyslipidemia[8], and physical activity has a direct effect on the movement of nutrients into the disc[9].

The role of nutrition as the main initiator of DD has been questioned since genetic and early environmental factors accounted for most of the variance in DD, while environmental factors played only a negligible role[10, 11]. However, almost all studies on the effect of environmental factors on DD have been performed in adult populations, and it is possible that the effect of environmental factors is more visible at a young age.

Our study population provides an excellent opportunity to assess the role of environmental factors in DD, as the cohort members have been followed up since birth. In the present study, we evaluated the effect of persistent smoking, body weight and physical activity at the ages of 16 and 19 years on lumbar DD in lumbar magnetic resonance imaging (MRI) performed at a mean age of 21.

High BMI at 16 and 19 years associated with lumbar DD at 21 years among males belonging to the birth cohort. Having smoked at least four pack-years was also associated with lumbar DD among males, even though the finding did not reach statistical significance at 5% level. Such associations were not observed among females. The level of physical activity was not related with DD in either gender.

The first degenerative changes such as increased cell death, cleft and radial tear formation, and cracks in the endplates can be seen already at 11 years of age at the time when the discs start to become avascular as the vessels penetrating into the disc through the endplates are obliterated[17]. At molecular level, DD is characterized by loss of proteoglycans, which leads to desiccation[18, 19]. Desiccation can be seen in MRI as the decreased signal intensity of the nucleus pulposus. MRI studies, in accordance with histological ones, have found a high prevalence of DD already in adolescence or early adulthood[11, 19–21] as we have earlier observed in our study population[12].

We are aware of only two previous studies on the association of DD with environmental factors among adolescents or young adults[22, 23]. Both studies were cross-sectional in design. The Japanese study[23], consisting of 308 university athletes and 70 non-athlete university students, observed that competitive baseball and swimming was associated with DD. The authors did not, however, evaluate the role of overweight or smoking. Among 13–20-year-old Southern Chinese subjects (n = 83), overweight or obesity based on Asian-modified BMI was significantly associated with the severity of lumbar DD, while smoking was not[24]. Unfortunately, their sample size was too small to allow gender stratification.

In the current study, BMI was associated with DD among males. We used two time points, 16 and 19 years, to study this association. Our results are consistent with a prospective study among Finnish male adults (n = 129) in which persistent overweight (BMI ≥25 kg/m2) at 25 and 40–45 years was associated with DD at L2/3 to L4/5 in MRI. Moreover, overweight at 25 years was associated with incident DD over a four-year period while overweight at 40–45 years was not[25]. Another longitudinal study[26] used the radiographic assessment of Dutch women’s osteophytes and narrowing of the disc space over a nine-year period (n = 742; age range 45–64 years at baseline). BMI was a predictor of DD only among those without DD at baseline. Cross-sectional designs have found associations between overweight and DD, both positive[23, 24] and negative[27]. Furthermore, BMI may be a risk factor of specific phenotypes of DD such as spondylosis and Modic changes[28–31]. Interestingly, among weight-disconcordant twins, overweight was associated with lesser desiccation of the intervertebral discs suggested to be due to slow adaption to mechanical loading[32]. The results were disputed by the findings of a British twin study in which weight-related factors were associated with DD[33].

A high level of physical activity is beneficial for metabolic syndrome and cardiovascular health[34]. Yet a high level of physical activity, especially participation in power sports, appears to accelerate the development of DD[23, 35–37]. Indeed, this topic remains controversial as some studies have found a positive association between heavy physical loading, either in occupation or sports, and DD, but negative findings have also been published[38–40]. The harmful effects of sports may be due to trauma, or excessive physical activity causing abnormal stress on the structural components of the disc with failure of the motion segment[41, 42]. In the current study, the overall level of physical activity between the ages of 16 to 19 years was not associated with DD with any certainty. We evaluated overall level of physical activity as MET hours per week, which earlier studies have not done. METs take into account every physical activity level and therefore subjects who exercise actively even at a low level may have high MET scores. MET acts as a surrogate measure of overall physical activity level, not only participation in physical exercise and sports. We did not, however, evaluate the association between different sports and DD and therefore, we cannot rule out existence of risk sports with harmful effect on disc well-being.

We found an association between smoking and DD among young adult males, which attenuated only slightly in the multivariate analysis. Previous population-based studies have found no such trend[24, 38, 43]. Among smoking-disconcordant adult twins, smoking explained only 2% of the variance in lumbar DD, which can be regarded as a marginal effect[10, 44]. However, in experimental studies, smoking has indisputably caused degenerative changes in the intervertebral discs[6, 7, 45]. It may well be that the association between smoking and DD can be observed more easily in a young population. Interestingly, a recent meta-analysis found a stronger association between smoking and low back pain (LBP) among adolescents than adults[46].

In our study, BMI was significantly associated with DD among males, whereas physical activity was not. Overweight may cause or accelerate DD by a low-grade systemic inflammatory state caused by obesity[47, 48]. We cannot exclude gene-environmental factor interactions such as those shown for obesity and the COL9A3 gene[49]. The deleterious effect of overweight on the well-being of intervertebral discs is supported by a recent study of surgical tissue samples, which showed that overweight correlated with histological degenerative abnormalities[50].

The clinical relevance of DD is questioned by the fact that its prevalence is high among asymptomatic subjects[51, 52]. However, increasing evidence has emerged that DD is associated with LBP both at a young age[12, 20, 22, 52, 53] and in adulthood[33, 54–56]. Large prospective population-based cohorts are needed to evaluate whether weight control or smoking cessation protect from low back pain.

The strength of our study is the population-based birth cohort design, in which many environmental factors have been assessed prospectively from birth onwards. For this study, we used data regarding persistent smoking, BMI and physical activity at 16 and 19 years. Anthropometrics (weight, height, WC) were based on measured values in most cases. BMI was calculated using measured height and weight, but we did not use the published cut-off values of obesity and underweight for children and adolescents[57, 58], because only a few subjects were outside the limits of the cut-offs. Moreover, using measurements at 16 and 19 years allowed us to evaluate the impact of persistence, in anthropometric measures. An additional strength of our study is the narrow age range, which enabled us to minimize the confounding effect of age.

The main limitation of our study is the cross-sectional design of imaging, which prevents us from drawing conclusions about temporal patterns between environmental factors and MRI findings due to the lack of sequential MRI scans. Therefore, the onset and progress of DD in the lumbar spine among the study subjects remains unknown. However, as DD has early onset[22], a very large cohort starting at early years with annual imaging for decades would be needed in order to study the natural progression of degenerative changes and their association with unhealthy behaviors.

A further limitation is that the data on smoking consisted of self-reported values. We used pack-years at 19 years and for physical activity the mean of MET hours per week at 16 and 19 years. We assume that young adults are more likely to underreport smoking than to overestimate it and, hence we regard the association between smoking and DD among males plausible. Assessment of physical activity was also based on self-reported values, which may have led to over-reporting of physical activity[59]. Unfortunately, more objective methods for assessment of physical activity such as pedometers, accelometers and heart rate monitors were not available in the present study. However, the test-retest reliability of physical activity questionnaire has been shown to be good at 16-year old population[14]. Correlation between self rated brisk physical activity and measured aerobic fitness has earlier been found to be acceptable also among 8 and 10 year old schoolchildren[60]. Another limitation of our study was that DD was evaluated visually from MRI. We used qualitative Pfirrmann classification[13], which is considered inferior to the quantitative assessment of DD[61, 62]. We agree that visual evaluation is more robust than quantitative assessment but we have earlier found a moderate-to-good inter-rater reliability art the three lowest levels[12]. The kappa-values were lower than previously reported[63], but the disagreements were settled by consensus in our study. Moreover, due to the young age of our participants, the influence of environmental factors on lumbar DD might be less than genetic factors. However, associations of environmental factors with DD clearly existed among our study participants.

The MRI participants (n = 563) were more likely females, sat for shorter times, were more physically active, were more likely non-smokers and more likely suffered from LBP than non-participants to MRI of the original OBS postal survey (n = 2408). Thus, although some participation bias occurred, its impact on the generalizability of the results is likely to be small since the differences were minimal. Furthermore, no relevant bias was observed between the participants and nonparticipants among the subjects invited to MRI, as only BMI was found to be somewhat lower among the scanned subjects[12].

Our study showed an association between BMI and DD among young males, whereas no such associations were observed among females. Also having smoked at least four pack-years showed a comparable association in males, which however failed to reach the 5% statistical significance. Physical activity had no significant association with DD in either gender. These results suggest that environmental factors are associated with DD among young males.