We found that pre-diabetes was associated with worsened levels of several major CVD risk factors as well as with increased carotid and femoral IMT independently of markers of adiposity (waist and BMI) or a marker of insulin resistance (insulin).
The smaller than expected effect of waist/BMI and insulin in explaining the relationships between IGR and the major CVD risk factors deserves several comments. First, we observed that BMI, waist or insulin attenuated the relationship between IGR categories to a fairly similar small extent and that adjusting for these factors altogether did not add substantial adjustment as compared to adjustment based on any of the three markers alone. This suggests that these three markers represent a same dimension/mechanism (e.g. insulin resistance) and that none of these markers conveys a substantial advantage in representing this dimension. In our results, waist tended however to perform slightly better than BMI. An advantage of waist over BMI has been found in some studies , but not all . The fairly similar non-additive effect of the three considered markers (insulin, BMI, waist) has practical clinical relevance: BMI (or waist) is much simpler and less expensive to asses as compared to insulin and BMI (or waist) might be preferred to insulin for risk stratification, particularly in resource-constrained settings.
Several factors may underlie the smaller than expected effect of the considered markers (BMI, waist, insulin) on the relationship between IGR and major CVD risk factors. First, this may reflect the known fact that only a sub-group of obese persons are insulin resistant and are at risk for developing IGR  and that obese individuals without insulin resistance have only marginally increased CVD risk [35, 36]. Second, BMI, waist and insulin may only imperfectly represent insulin resistance and the effect of such markers on the relationship between IGR and major CVD risk factors might have been larger, had we used better indicators of insulin resistance. However, better markers of insulin resistance, such as the euglycemic clamp, would require complex measurements that are not practical for epidemiological studies or usual clinical practice. Third, insulin resistance is a broad description underlying many different altered physiological factors. In particular, insulin resistance has been associated with a variety of pathophysiological effects related to abdominal fat, including cytokines secretion and inflammatory cell migration . For instance, the proinflammatory cytokine TNF-alpha may impair intracellular insulin signaling independently of insulin . BMI, waist and insulin may therefore only poorly correlate with such finer physiological factors. In particular, insulin may not be a reliable marker for insulin resistance and subsequent atherosclerosis . This could particularly be the case in diabetic persons with depleted insulin secretion (whether type 1 or type 2 DM). However, a moderate effect of adiposity/insulin in explaining the relationship between IGR and major CVD risk factors is at odds with trials showing decreased incidence of DM among pre-diabetic persons who decreased their weight through lifestyle interventions  or through bariatric surgery . A possible explanation underlying this contradiction is that these interventions not only decreased fat mass or insulin but also improved other factors which may directly improve insulin resistance, e.g. physical activity, nutritional patterns or secretion/action of several hormones (e.g. incretins).
Similarly to the association with major CVD risk factors, the relationship between IGR and IMT was only partially explained by waist, BMI and insulin. In other studies, carotid IMT was associated with 2hBG but with neither fasting glycemia nor a insulin sensitivity index [41, 42]. Consistent with our results, IMT remained significantly associated with 2hBG upon adjustment for major CVD risk factors, waist, BMI and insulin sensitivity index . Interestingly, we found that the relationships between IGR and IMT tended to be larger at femoral than carotid levels independently of adjustment. We are not aware of any other study that has investigated these associations between IGR and IMT at femoral level. However it was recently shown that metabolic syndrome components impacted selectively on IMT at the femoral site: insulin and triglyceride concentrations were strongly associated with femoral IMT but not with carotid IMT . These findings suggest the usefulness of femoral IMT for assessing CVD outcomes [18, 43].
In addition, the relationships between IGR and either major CVD risk factors or IMT may be linked to several specific mechanisms. It is shown that atherosclerosis is accelerated by insulin resistance and DM . Atherosclerosis, which also encompasses an inflammatory process , is in turn related to several factors associated with adiposity and/or insulin resistance. These factors include numerous adipokines  that are released or modulated by adipose tissue. From another perspective, insulin resistance has also been shown to alter the endothelial function, which can result in impaired production of NO , reduced blood flow, pro-inflammatory state and pro-thrombotic state [46, 47]. Furthermore, hyperglycemia can also induce atherosclerosis independently of insulin, e.g. through glycation of proteins and lipids and by increasing oxidative stress .
We found that IGT was associated with major CVD risk factors or IMT more strongly than IFG. This different significance of IFG and IGT is consistent with other studies showing a stronger association of IGT than IFG with CVD risk factors  and with IMT [41, 49], but not with a recent prospective study linking CVD and total mortality at least as strongly with IFG as with IGT . A greater impact of IGT than IFG may relate to the facts that subjects with IFG and IGT are more likely to be insulin resistant whereas subjects with IFG and NGT are more likely to have insufficient insulin secretion [48, 50, 51]. It has been suggested that a pro-inflammatory state leading to CVD seems restricted to those individuals with IGR who are insulin resistant, measured by an insulin sensitivity index  or HOMA-IR .
Several limitations of this study need do be considered. First, we did not perform OGTT in individuals with FBG < 5.6 mmol/l and we could have missed a few cases of DM and IGT in persons with NFG but pathologically high 2hBG. The number of such cases is however expected to be small . A second limitation is related to the cross-sectional nature of our study which precludes defining causal relationships. Third, we may not have captured insulin resistance optimally with BMI, waist and insulin. insulin was shown to relate only moderately with insulin resistance measured by euglycemic clamp . Also, BMI and waist are only proxy measures of total adiposity and intra-abdominal adiposity. Yet, this study adds to the limited information on the associations between pre diabetes, on the one hand, and CVD risk factors and peripheral artery IMT, on the other hand. We are not aware of any previous study that has examined this issue in a population in the African region.