Obesity was associated with a decline in GFR estimated by the MDRD Study equation over 7 years of follow-up in a large, community-based population of older adults. Results were consistent within each sex and were observed whether obesity was measured as BMI, waist circumference, or fat mass using bioelectrical impedance. Associations of obesity with GFR loss were stronger among participants with impaired GFR at baseline and were substantially attenuated by adjustment for diabetes, hypertension, and C-reactive protein.
These results are consistent with observations in younger cohorts, which have often but not always observed associations of obesity with impaired kidney function.4–11, 32
Our focus on older adults is particularly important because: (1) the prevalence of CKD is highest and increasing most rapidly among older adults; (2) risk factors for kidney disease in older adults are poorly understood; (3) associations of obesity with non-renal health outcomes have not been consistently found in older populations; and, (4) kidney disease appears to confer a particularly poor prognosis for cardiovascular and mortality outcomes among older adults.1–3, 14
Our results suggest that obesity may be a modifiable risk factor for the development and progression of kidney disease among older adults.
In contrast to results assessing MDRD Study estimated GFR as an outcome, obesity measurements were not associated with change in GFR estimated using serum cystatin C. Reasons for this discrepancy are not clear and require further investigation. However, our data raise the possibility that, among participants who were more obese at baseline, loss of renal cystatin C elimination over follow-up was balanced by simultaneous loss of cystatin C production (which derives in part from adipose tissue), resulting in no observed net change in steady-state serum cystatin C concentration. This hypothesis is supported by the observed inverse correlations of obesity measurements both with baseline GFR estimated from cystatin C (cross-sectional correlations, ) and with change in BMI over follow-up. In contrast, serum creatinine is not produced by adipose tissue, so that changes in serum creatinine caused by loss of renal function are less likely to be obscured by simultaneous loss of adipose tissue. Supporting this concept is the published observation that the relationship of change in MDRD Study estimated GFR with change in iothalamate GFR was not biased by baseline BMI in the MDRD Study, though the MDRD Study contained relatively few obese individuals.33
Thus, production of endogenous filtration markers is an important consideration when assessing the relationship of obesity with kidney function,34–36
and this may be a unique circumstance in which creatinine-based GFR estimates have advantage over GFR estimates based on serum cystatin C.
Associations of obesity with loss of MDRD Study estimated GFR were strongest when baseline GFR was less than 60 mL/min/1.73m2. It is possible that obesity places a physiologic strain on the kidney that is most detrimental in the setting of prior kidney damage and limited renal reserve. In this setting, obesity may act as a “second hit,” accelerating the progression of CKD. Alternatively, the detrimental effects of obesity on GFR loss may be equally strong in persons with normal baseline GFR, but the strength of association in this subgroup may be underestimated due to imprecise GFR estimation in the normal range.
Associations of obesity with GFR loss were substantially attenuated by adjustment for diabetes, hypertension, and C-reactive protein (causal pathway model). This is consistent with prior studies and suggests that obesity may adversely affect kidney function in large part through established mechanistic pathways: impaired glucose metabolism, blood pressure, and inflammation. Alternatively, diabetes, hypertension, and elevated C-reactive protein may reflect underlying poor health which leads to both obesity and GFR loss, thus confounding the association of obesity with GFR loss.
Strengths of this study include its focus on older adults, multiple measurements of obesity and adiposity, and longitudinal measurement of estimated GFR to assess change over time. In addition, we avoided use of a threshold level of GFR to define impaired kidney function. Use of thresholds may introduce bias when evaluating obesity as a risk factor because the same estimated GFR may reflect higher true GFR for people with greater body size due to increased creatinine/cystatin production.9, 37
There are also potential limitations to this study. First, results are susceptible to survival bias, as with all studies requiring repeated measurements, because participants with fewer than two kidney function measurements were excluded. Fortunately, excluded patients did not differ substantially from those included in analyses with regard to baseline obesity measurements. Second, direct measurements of GFR were not available. In particular, use of GFR estimates may have underestimated true loss of GFR in the cohort,33
and differences comparing GFR estimates based on creatinine versus cystatin cannot be conclusively explained. Third, results may not apply to younger populations, or to populations with more severe CKD.
In conclusion, obesity is associated with increased risk of GFR loss in older adults, when GFR is estimated using serum creatinine. This relationship appears to be strongest among persons with impaired GFR at baseline and to be largely the result of higher prevalences of diabetes, hypertension, and inflammation among more obese individuals.