Atherosclerotic cardiovascular disease (CVD) is the major cause of death in patients with type 1 diabetes mellitus (T1DM). Alterations in the HDL proteome associate with prevalent CVD in T1DM. We therefore determined which proteins carried by HDL might predict incident CVD in T1DM patients. Using targeted MS/MS, we quantified 50 proteins in HDL from 181 T1DM subjects enrolled in the prospective Coronary Artery Calcification in Type 1 Diabetes study (CACTI). We used Cox proportional regression analysis and a case-cohort design to test associations of HDL proteins with incident CVD (myocardial infarction, coronary artery bypass grafting, angioplasty, or death from coronary heart disease). Only one HDL protein—SFTPB (pulmonary surfactant protein B)—predicted incident CVD in all of the models tested. In a fully adjusted model that controlled for lipids and other risk factors, the hazard ratio was 2.17 per SD increase of SFTPB (95% confidence interval, 1.12-4.21, P=0.022). Plasma fractionation demonstrated that plasma SFTPB is nearly quantitatively bound to HDL. Although previous studies have shown that high plasma levels of SFTPB associate with prevalent atherosclerosis only in smokers, we found that SFTPB predicted incident CVD in T1DM independently of smoking status. Elevated levels of SFTPB in HDL strongly predicted incident CVD in CACTI subjects independently of a wide range of confounding factors, including smoking status, HDL-C, LDL-C, and triglyceride levels. Because SFTPB is almost quantitatively bound to plasma HDL, our observations support the proposal that SFTPB carried by HDL is a marker—and perhaps mediator—of CVD risk in patients with T1DM.

To quantitatively measure the relative levels of HDL proteins, we used targeted proteomics with isotope-dilution parallel reaction monitoring (PRM). PRM analyses were performed in the positive ion mode with an ultrahigh-resolution accurate mass Orbitrap Fusion Lumos Mass Spectrometer coupled to a nanoACQUITY UPLC. Initially, the potential peptides for each protein were selected from the detected peptides by shotgun analysis and from our previous studies. At least two peptides from one protein were then tested by PRM test runs and finally 2 or more peptides were selected for 35 proteins, and 1 peptide for 15 proteins. The targeted PRM MS data of peptides of each HDL protein were analyzed using Skyline. An equal amount of 15N-labeled APOA1 was added to HDL isolated from each subject prior to digestion as an internal standard. The peak areas of all the transitions of a peptide detected by PRM analysis were summed to get the total peak area for the peptide but the transitions with interferences were deleted. To normalize the peak area of a peptide, the total peak area of all selected transitions of the peptide was divided by the peak area of one 15N-labeled peptide (DYVSQFEGSALGK) from 15N-APOA1 and the ratio was used for quantification. To calculate the relative levels of the peptide between control and incident CVD groups, we set each of the average ratio of the peptide in control subjects as an arbitrary unit of one and the average of the four ratios was set as the relative level of the peptide. If two peptides were quantified for a protein, the relative levels of all peptides from the protein were averaged to obtain the relative level of that protein in HDL.

We used plasma samples from the CACTI sub-cohort (134 control subjects and 11 subjects with CVD events) and all other subjects (N=36) with CVD events outside the sub-cohort that were collected at the time of enrolment. HDL (density 1.063-1.210 g/mL) was isolated by sequential ultracentrifugation from rapidly thawed plasma. 8 µg of HDL proteins and 0.4 µg of isotope-labeled [15N]APOA1 (added as the internal standard) were digested by trypsin. And the digested peptides were analyzed to determine the relative levels of 50 proteins in HDL.