This study demonstrates that in diabetic patients, non-HDL-C and apolipoprotein B100 performed equally well to discriminate patients according to their atherogenic cholesterol values or atherogenic particles number. In addition, as the underlying correlation between these two continuous variables reached unity once attenuation was taken into account, these two measurements may be used to assess what represents an equivalent underlying biological condition, and that they can substitute for each other for that respect.
While it is well-recognized that non-HDL-C and apoB are closely related metabolically, yet there is ongoing discussion as to whether one should be measured preferentially over the other, with some considering apoB as a choice proatherogenic index in patients with cardiometabolic risk associated with atherogenic dyslipidemia. Our results confirm the equivalence of both measurements in their capacity to rank diabetic patients with a broad spectrum of lipid values, from normal to frank dyslipidemia with elevated atherogenic cholesterol (non-HDL-C) and/or atherogenic particles number (apoB).
In this study involving diabetic patients, apoB did not perform significantly better than non-HDL-C to rank patients according to atherogenic dyslipidemia. Noteworthy, lack of significant difference in DRs was not a bias caused by accretion dilution from pooling T1DM and T2DM patients with divergent DRs. Whereas non-HDL-C is derived from the comput of two robust, well-established measurements methods (total cholesterol and HDL-C), its DR may be negatively affected by amplification error from the incorporation of the intrinsic imprecisions of two measurements. The small difference (0.07) in DR that we observed in this study may be viewed as not clinically meaningful, and there were no significant differences in discrimination between non-HDL-C and apoB in both T1DM and T2DM subgroups, as well as in the combined group.
Such difference in the DR of non-HDL-C and that of apoB would nevertheless translate into a meaningful statistical superiority in the setting of much larger cohort sizes (i.e. n > 3500 when dealing with duplicate measurements or n > 1800 for triplicates). Thus, the higher DR of apoB may represent one contributor to higher performance of determining atherogenic particles number (apoB) against measuring atherogenic cholesterol level (non-HDL-C) to predict coronary heart disease, as observed by Pischon et al.. In a routine clinical setting, usually dealing with individual risk estimation or risk stratification between dozens to hundreds of patients, both non-HDL-C and apoB performed equally well to discriminate between diabetic patients, and may be considered as interchangeable surrogates in diabetes.
In the long run, apoB may eventually supersede non-HDL-C, once a sufficient number of large prospective studies confirm its superiority in risk prediction, and after a consensus is reached regarding between-assay standardization. ApoB determination alleviates the requirement for fasting conditions, as mentioned in INTERHEART . Meanwhile, the DR method allows establishing an unbiased equation of equivalence relating non-HDL-C to apoB in diabetic subjects. Since the former is readily available from routine lipid profile in the fasting state whatever the TG level, this equation alleviates the need to perform additional, and at present more costly, apoB determination. Once attenuation was taken into account, both measurements were related by a line of unity, and had similar performance for ranking diabetic subjects according to the risk afforded by apoB-containing particles in scope with the current cholesterol hypothesis paradigm. As both non-HDL-C and apoB represent essentially the same underlying biological variables in diabetics, both can therefore be substituted to each other once an unbiased equation of equivalence is used.
This study has several potential limitations. An easily refutable one is sample size, as the DR methodology does not require large samples, with n ≥ 20 for 2 replicates being adequate as long as sample represents patients spread over a meaningful clinical range for the variables under study . As the majority of T1DM are normolipaemic, combining T1DM and T2DM patients with significant differences in both non-HDL-C and apoB, as well as patients with and without lipid-lowering drug(s), generated an optimal spread of atherogenic lipids and particles number, as required by the DR methodology for assessing the performance of a continuous physiological variable. In contrast, the reported observations need to be verified in other non-diabetic populations with various severities of (un)treated dyslipidemia. A second limitation is that postprandial lipids were not compared, and the conclusions may not necessarily apply to this state. Yet, neither non-HDL-C nor apolipoprotein B100 levels are substantially affected by postprandial excursions in TG-rich lipoproteins, the latter arising mostly from de novo secretion of gut-derived apolipoprotein B48-containing chylomicrons.
Although non-HDL-C is derived from two separate biological assays (total cholesterol and HDL-C) each with respective imprecision and day-to-day variation, the DR methodology demonstrates that the discrimination of non-HDL-C was similar to that of apoB, measured from a single assay. Non-HDL-C translates into biological measurement of both the number (since it mostly estimates cholesterol from apoB-carrying lipoproteins) and the combined cholesterol mass of atherogenic apoB-containing particles. Non-HDL-C determination includes cholesterol from atherogenic lipoproteins not captured by Friedewald's estimation, such as apoB-carrying remnants not belonging to LDL or IDL and often found in diabetic patients and/or in subjects with metabolic syndrome [14, 16, 18]. The high agreement that we observed between methods in this population may also reflect bidirectional shifts in the distribution of apoB-containing particles size and cholesterol content seen in diabetic states, including concomitant increases in TG-poor (such as small-dense LDL), as well as TG-rich particles.
In parallel to determination of circulating apolipoproteins, lipid ratios, still widely reported on lab reports, incorporate measurements of total or atherogenic lipids (total C, LDL-C, apoB as numerator) and measurements of HDL-C or apoA-I to evaluate reverse cholesterol transport particles as denominator. We previously reported that with respect to ratio-based lipid markers, the highest DR's were those of total C/HDL-C and non-HDL-C/HDL-C, which were significantly better than estimated LDL-C/HDL-C. In addition, the apoB/apoA-I ratio had a non-significantly higher DR than non-HDL-C and estimated LDL-C/HDL-C . Due to physiopathological relevance, robustness, ease and low-cost, and imperviousness to fasting conditions, the discriminant ratio of non-HDL-C/HDL-C renders it suitable for routine ranking of atherogenicity in a given patient with type 2 diabetes and/or insulin resistance, exposed to high number of LDL (including small and dense ones), VLDL and remnants lipoproteins.
With respect to LDL-C and non-LDL-C therapeutic targets, a recent joint Consensus Statement (American Diabetes Association and American College of Cardiology Foundation) suggests new treatment goals for apoB in patients with atherogenic dyslipidemia and cardiometabolic risk. An apoB level <90 mg/dL was proposed in patients without diabetes or known CVD but with ≥2 additional major CVD risk factors, or with diabetes and without major CVD risk factors. An apoB level <80 mg/dL was suggested for patients with the highest CVD risk, i.e. known CVD or diabetes plus ≥1 additional major CVD risk factor . With respect to other determinations, there is growing evidence suggesting that direct measurements of other specific apolipoproteins may contribute to refine CV risk assessment. Thus, apolipoprotein CIII (apoCIII) level may capture a sizeable component of TG-attributable CV risk, and apoCIII measurement may be suited to estimate aspects of hepatic production of a highly-atherogenic VLDL subset and of their derived atherogenic particles, although at the moment apoCIII measurement remains in the realm of research [23–25].
In conclusion, the present study demonstrates, using the validated DR methodology, that the discrimination of non-HDL-C is similar to that of apoB in diabetic patients. Non-HDL-C represents not only a metabolic surrogate, but is close to a true biological equivalent of apoB in this specific population. Besides its original usefulness for estimating LDL-C atherogenicity in hypertriglyceridemic patients with TG values outside of Friedewald's formula's range, non-HDL-C is an easy and cost-effective means to estimate apoB levels, while waiting for a consensus whether to use apoB (i) as an alternative to LDL-C for biological assessment of hypercholesterolemia, (ii) as a residual risk assessment tool in dyslipidemia, or (iii) as secondary therapeutic target beyond routine lipids measurement [26–28].