Upsala J Med Sci 81: 71-78, 1976 Lipoprotein Composition and Lipoprotein Interrelations in 50-year-old Men with Hyperlipoproteinaemia BENGT VESSBY, HANS HEDSTRAND and U L F OLSSON From the Departments of Geriatrics, lnternal Medicine, and Statistics, University of Uppsala, Sweden Al3 STRAC T The serum lipoprotein (LP) composition and LP lipid inter- relations were studied in 50-year-old men with different types of hyperlipoproteinaemia (HLP) and in randomly sampled healthy controls from the same population. The ratio cholesterol/triglycerides in very low density lipoproteins (VLDL) was high in HLP type 111. The other types of HLP showed ratios not significantly different from the controls. The low density Lipoprotein (LDL) cholesterol concentration was similar in controls, type 111 and type IV while, by definition, higher values were seen in type I I A and IIB. All types of HLP showed statistically significantly higher LDL triglycerides than the controls. HLP type I1 A and I1 B showed cholesterol/triglyceride ratios in LDL similar to the controls. The corresponding ratio in type I V was lower than in the control subjects but the lowest ratio was seen in type 111 with a mean value below the 5th per- centile of healthy controls. The high density lipoprotein (HDL) cholesterol con- centration was decreased in HLP type IV. Apparently elevated HDL triglyceride levels were seen in all types of HLP with the highest mean value in type 111. The LP Lipid interclass relationships were analysed in the random sample of healthy men and compared to cor- responding relationships in the different types of HLP. Apart from HLP type III and HLP type I V with low LDL cholesterol levels all other types of LP patterns seemed to conform to a common model of LP interconversions. In type IV a significant negative correlation between VLDL concentration and LDL cholesterol levels was demonstrated in subjects with low LDL cholesterol as well as a direct relationship between LDL cholesterol concentration and the cholesterol/triglyceride ratio in VLDL. There were no significant correlations between LDL cholesterol concentra- tion and VLDL lipid variables in other types of HLP and normolipidaemia. INTRODUCTION Hyperlipoproteinaemia (HLP) is one of the major risk factors for development of premature athero- sclerotic cardiovascular disease (14, 3). According to suggestions by Fredrickson et al. (7, 8) H L P are divided into separate ‘types’. The classification of H L P is based on the definition of upper ‘normal’ limits for the concentration of very low density lipoproteins (VLDL, d< 1.006) and low density lipoproteins (LDL, d= 1.006-1.063) (1). The pre- valence of H L P in a population, and also the rela- tive frequency of the different types of H L P (13), thus depends on which cutting points are chosen. Lipid levels vary with sex and age as well as ethnic and geographic factors. This study was undertaken to characterize lipoprotein (LP) com- position and L P lipid interrelations in different types of H L P in 50-year-old men. The lipid levels in the isolated L P density fractions and the statisti- cal interrelationships between the LP fractions were studied and compared to corresponding rela- tions in randomly sampled healthy controls. MATERIAL AND METHODS Subjects During 1971-74 all men aged SO (born 1921-24) living in and around the town of Uppsala in eastern Sweden were invited to a health screening for risk factors for coronary heart disease. The adherance rate was 84%. All men with triglyceride and/or cholesterol concentration in serum in the two top decentiles of the population at the initial screening were referred to the Department of Geriatrics for a complete LP analysis. The LP patterns were classi- fied according to Fredrickson, Levy & Lees (8). Cutting points for HLP were for VLDL triglycerides and LDL cholesterol 1.5 mmol/l and 190 mg/lOO ml respectively corresponding to the 85th percentile in a randomly sampled population of healthy 50-year-old men (13). I n the present study all samples showing a characteristic LP pattern diagnostic for HLP according to WHO ( 1 ) ana- lysed from January 2,1972 through January 31,1973 were included ( n = l l l ) . Diagnosis of HLP type I11 was based on the demonstration of a typical ‘floating @’ pattern in VLDL.’ No samples from subjects suffering from insulin- * @-migrating lipoproteins demonstrated at agarose electrophoresis in the supernatant fraction of plasma after ultracentrifugation at d = 1.006. Upsala J Med Sci 81 7 2 B . Vessby et al. Table I. Triglyceride and cholesterol concentration in serum lipoprotein density fractions in 50-year-old men with normal lipoprotein pattern and hyperlipoproteinaemia (Mean fS.E.) VLDL" LDL HDL Lipoprotein Triglyceride Cholesterol Triglyceride Cholesterol Triglyceride Cholesterol pattern (n) ( m m o ~ / ~ ) (mg/100 ml) (mmol/l) (mg/IW ml) (mmol/l) (mg1100 mi) Controls (92) 1.03f0.06 I1 A (38) 1.02f0.05 *** I1 B (28) 2.12+0.11 * * * 111 (8) 2.18f0.26 *** IV (37) 2.85f0.38 21f 1 23+ 2 *** 41+ 2 *** 9 9 f 15 *** 532 6 0.49f0.01 155f 3 *** *** 0.68f0.03 224f 8 *** *** 0.70f0.02 219f 4 *** 0.71fO. 10 143f20 *** 0.61f0.03 145f 5 0.24f0.01 48f 1 * * 0.27f0.01 4 6 f 2 *** 0.29f0.01 4 3 f l *** 0.34f0.04 4453 *** *** 0.30f0.01 36f I a The tests involving VLDL lipid concentrations were performed also on logarithmic transformed values with identical results regarding the demonstrated group differences. * 9 , ** ***=Significant on the 2, 1, and 0.1% level respectively compared with normals when tested with Student's t-test. deficient diabetes, hypothyreosis or renal disease were in- cluded. One man had suffered from a myocardial infarc- tion. One had angina pectoris and was treated with a /3 receptor blocking drug and one suffered from intermittent claudication. N o other subjects were on regular treat- ment. Seventeen men had a reduced glucose tolerance at intravenous glucose tolerance test. Three of those showed a moderately increased fasting blood glucose concentra- tion without glucosuria. Controls The control subjects (n=92) were apparently healthy men randomly sampled from the same population of 50-year- old men. Excluded from this material were obese persons (subjects with weightlheight index above 1.10) and per- sons with clinical or laboratory signs of disease. The con- trol population has earlier been described in detail (13). LP analysis Blood samples from subjects fasted over night were al- lowed to clot at room temperature and EDTA was added as a 5 % solution to a final concentration of 0.05%. VLDL, LDL and high density LP (HDL, d>1.063) were isolated by consecutive spins at 15°C in a L2 65B Beckman preparative ultracentrifuge according to Have1 et al. (1 1) using a 40.3 rotor. VLDL was isolated as the top frac- tion after centrifugation of serum at d=1.006 for 16 h at 40000 rpm. The bottom fraction was then centrifuged at d=1.063 at 40000 rpm for 20 h. The top and bottom fraction after the second centrifugation contained LDL and HDL respectively. A detailed description of the isola- tion procedure has been given (2). Whole serum as well as the isolated L P classes were extracted manually with isopropanol. Triglyceride and cholesterol concentrations in the L P fEactions were determined in a Technicon Auto Analyzer Type I1 (23). The sum of cholesterol and TG concentrations in VLDL, LDL and HDL, was always within 100fIO% of whole serum cholesterol and tri- glyceride concentrations respectively. Immediately after the centrifugation whole serum and the top and bottom fraction at d= 1.006 were subjected to agarose electrophoresis according to Noble (19). A 1 % agarose gel containing 0.25% albumin was used. The electrophoresis was run in a barbital buffer, pH 8.6, at 16 V/cm for 1 hour and the gel was stained in Sudan Black. Statistics Significant differences between mean values were estimat- ed with Student's t-test (two-tailed tests). VLDL tri- glyceride and cholesterol concentrations as well as LDL triglyceride concentrations were tested also after loga- rithmic transformation because of a skewed distribution of these variables (13). Correlation analyses ( 5 ) were per- formed at the Uppsala University Data Center on an IBM 370/155 computer using program BMD02R (4). The accepted level for statistical significance was p<0.02. A significance at the 5 % level would be expected to occur by chance alone in one test of 20. Because of the number of significance tests performed (e.g. Table I and IV-VI) the minimum requirement for statistical significance was set at 2% to reduce the risk of chance significances. Since the different groups of H L P were obtained by selection, i.e. by assigning subjects with values above and below certain limits for some variables to different groups, there is a certain accordance between the group compari- sons of mean values and the reported correlation coeffi- cients. By definition, significant group differences were obtained for the group assignment variables. Similarly, group differences should be expected for variables highly correlated with the group assignment variables. Upsala J Med S c i S l Lipoprotein composition in 50-year-old men with hyperlipoproteinaemia 73 Table 11. Cholesterolltriglyceride ratio (Mean kS.E.1 in serum lipoprotein density fractions in 50-year-old men with normal lipoprotein pattern and hyperlipoproteinaemia - VLDL" LDL HDL Lipo-, mgll00 ml mgllOO ml mgllOO ml protein pattern ( n ) mmol/l mmol/l mrnol/l Controls (92) 20.4f0.6 325f 7 2 2 1 f 9 *** I I A (38) 22.2k1.2 340211 172? 8 *** IIB (28) 19.9k0.8 317f 8 154k 7 *** *** 111 (8) 44.8k3.6 2 0 2 t 1 0 1 5 5 f 3 9 *** * * * IV (37) 19.3k0.7 2 5 1 f l l 1 2 8 5 7 The tests involving VLDL lipid concentrations were performed also on logarithmic transformed values with identical results regarding the demonstrated group dif- ferences. *, **, ***=Significant on the 2, 1, and 0 . 1 % level rspectively compard with normals when tested with Student's t-test. RESULTS By definition H L P type IV and I I B had increased VLDL concentrations compared with the control subjects (Table I). The criteria for diagnosis of type 111 H L P do not include a requirement for increased lipid levels in VLDL o r LDL (1). However, on the average H L P type 111 showed increased VLDL tri- glyceride levels. The VLDL cholesterol concentra- tion was by far highest in type 111. The LDL cholesterol concentration was on the average similar in normals, type 111 and type IV. Significantly higher values were, by definition, seen in type I I A and IIB. All types of H L P showed sig- nificantly higher LDL triglycerides than the normal controls. Type 111, I I B and I I A showed the most pronounced rise while type IV showed LDL tri- glyceride levels intermediate between normals and the other types of H L P with a significantly lower mean value than in type I I B ( p < O . O l ) . Also in HDL the lipid levels differed in the various types of HLP. Type I I A showed HDL cholesterol concentrations similar to the normals. Types I I B and 111 tended to show lower mean values than the controls but the differences were not statistically significant. In type IV there was a clearly lower HDL cholesterol concentration than in the control group. Apparently elevated HDL tri- glyceride concentrations were seen in all types of HLP. The highest mean value was seen in type 111. In an effort to gain further information regard- ing the lipid composition in different hyperlipo- proteinaemic states the ratios between cholesterol and triglyceride concentrations in VLDL, LDL and HDL were computed and compared with normals (Table 11). Although not significantly different there was a tendency to a higher VLDL ratio in I1 A and lower in type IV than in normals. Type 111 was characterized by a very high ratio in VLDL. H L P types I1 A and I I B showed cholesterol/tri- glyceride ratios in LDL similar to the controls. Type I I A showed the highest mean value although not significantly different from the two other groups. The ratio in type IV was on the average Table 111. Relationships between triglyceride ( x ) and cholesterol ( y ) concentration in serum lipoprotein in 50-year-old men with different lipoprotein patterns Linear correlation coefficients ( r ) , slope ( b ) and intercept ( a ) of regression equations (y=a + b x ) Lipoprotein pattern Lipo- Controls (n=92) 11 A (n=38) I I B (n=28) 111 ( n = 8 ) IV (n=37) protein fraction r b a r b a r b a r b a r b a *** *** *** * * *** * * VLDL 0.89 18.9 1.2 0.70 22.8 -0.3 0.68 12.6 14.6 0.84 50.6 -11.5 0.95 14.9 10.3 * * * *** ** * * * ** * * * *** ** *** LDL 0.60 169 72 0.49 140 129 0.52 98 150 0.94 193 7 0.44 72 101 *** *** *** ** *** HDL 0.10 21 44 0.23 46 33 -0.05 -8 46 0.04 4 43 -0.06 -7 38 * > , ** ***=Significantly different from zero on the 2, 1, and 0.1 % level respectively. Upsula J M e d Sci 81 74 B . Vessby et al. Table IV. Relationships between lipoprotein lipid variables in the very low and low density lipoproteins in 50-year-old men with different types of serum lipoprotein patterns Linear correlation coefficients ( r ) LDL Controls (n=92) IIA (n=38) I1 B (n = 28) IV (n=37) VLDL" TG Chol Chol/TG TG Chol Chol/TG TG Chol Chol/TG TG Chol Chol/TG * ** *** *** TG 0.26 0.06 -0.27 0.22 0.03 -0.21 -0.08 -0.08 0.06 0.05 -0.62 0.55 ** ** *** * * * * *** Chol 0.35 0.11 -0.30 0.52 0.08 -0.55 0.27 0.00 -0.28 0.33 -0.40 -0.68 *** *** * * * * ** Chol/TG 0.21 0.11 -0.10 0.59 0.06 -0.66 0.44 0.13 -0.41 0.62 0.43 -0.32 VLDL triglyceride and cholesterol concentrations were logarithmic transformed before calculations. * , 1 ** ***=Significant on the 2, 1, and 0.1 % level. clearly lower than the controls. The lowest ratios, however, were seen in type I11 where the mean ratio was well below the 5th percentile of randomly sampled healthy controls. I n H D L all types of H L P exhibited a mean value for the cholesterol/triglyceride ratio below the con- trols. Type I I A showed a moderately decreased ratio, in type IV the ratio was reduced with nearly 50%. Types I I B and I11 showed similar average ratios. The relation between cholesterol and triglycer- ides in the L P fractions were studied by correla- tion analysis in the different types of H L P and com- pared to those found in the population of randomly sampled healthy controls (Table 111). In the controls a strong positive correlation was seen between VLDL triglycerides and cholesterol (r=0.89). A less pronounced but clearly significant correlation was also seen in LDL (r=0.60). In HDL, however, there was no significant correlation between triglyc- erides and cholesterol (r=0.10). The same pattern was seen in all types of HLP: A significant positive correlation between cholesterol and triglycerides in VLDL and LDL, no correlation in HDL (Table 111). When LP lipid interclass relationships were analyzed in the control sample certain significant lipid relationships were revealed. These were basi- cally similar to those distinguished also in H L P types I I A , I I B and IV (Tables IV-VI). These cor- relations were not studied in H L P type 111 because of the low number of subjects in this group. The LDL triglyceride concentration tended to Table V. Relationships between lipoprotein lipid variables in the very low and high density lipoproteins in 50-year-old m e n with different types of serum lipoprotein patterns Linear correlation coefficients ( r ) HDL Controls (n =92) I I A (n=38) I1 B (n =28) IV (n=37) VLDL" TG Chol Chol/TG TG Chol CHol/TG TG Chol Chol/TG TG Chol Chol/TG *** *** *** ** * * *** *** *** TG 0.36 -0.41 -0.51 0.07 -0.43 -0.42 0.23 -0.01 -0.12 0.47 -0.52 -0.54 *** *** *** *** *** *** * * *** Chol 0.37 -0.46 -0.54 0.26 -0.54 -0.60 0.27 -0.04 -0.20 0.52 -0.42 -0.57 * * * Chol/TG 0.07 -0.14 -0.13 0.28 -0.41 -0.50 0.03 -0.03 -0.08 0.15 0.18 0.09 VLDL triglyceride and cholesterol concentrations were logarithmic transformed before calculations. * , , ** ***=Significant on the 2, 1, and 0.1 % level. Upsulu J Med Sci 81 Lipoprotein composition in 50-year-old men with hyperlipoproteinaemia 75 Table VI. Relationships between lipoprotein lipid variables in the low and high density lipoproteins in 50-year-old-men with different types of serum lipoprotein patterns Linear correlation coefficients ( r ) HDL Controls (n=92) I I A (n=38) IIB (n=28) IV (n=37) LDL TG Chol Chol/TG TG Chol Chol/TG TG Chol Chol/TG TG Chol Chol/TG ** ** * * * TG 0.28 -0.19 -0.30 0.28 -0.37 0.43 0.30 -0.10 -0.27 0.29 -0.09 -0.39 Chol 0.04 -0.04 -0.11 0.09 0.01 0.12 0.03 -0.11 -0.11 -0.22 0.30 0.17 * * *** * * * *** ** Chol/TG -0.28 0.20 0.22 -0.31 0.43 0.59 -0.34 0.01 0.22 -0.54 0.31 0.59 * * * ***=s. ignificant on the 2, 1, and 0.1 % level. , > vary positively with VLDL lipid levels (significant in controls and type I1 A) (Table IV) and was signi- ficantly positively correlated to the cholesterol/ triglyceride ratio in VLDL in both H L P types I1 A , IIB and IV. The LDL cholesterol/triglyceride ratio tended to vary inversely with VLDL lipid con- centration (significant in controls, types I1 A and IV) and with the VLDL ratio (significant in type IIA). The LDL cholesterol concentration on the other hand did not show any significant relation- ships with lipids in other LP classes in either controls, I1 A or I1 B with remarkably low cor- relation coefficients when related t o all other lipid variables. In type I V , however, a negative correla- tion to VLDL lipid concentrations as well as a positive correlation to the VLDL cholesterol/tri- glyceride ratio was demonstrated. The negative correlation to VLDL as well as the positive correla- tion to the VLDL ratio was due to the presence of a number of patients with H L P type IV with high VLDL triglyceride levels and low LDL cholesterol levels. HDL triglyceride and cholesterol concentrations generally tended to show a positive and negative relation to VLDL lipid concentration respectively (Table V). The HDL ratio was negatively correlat- ed to the VLDL lipid concentration. The H D L triglyceride concentration was positively respec- tively negatively correlated to LDL triglyceride levels and the LDL cholesterol/triglyceride ratio in the control groups (Table VI). The HDL choles- terol concentration in type I I A and the H D L tri- glyceride concentration in type IV were positively respectively negatively correlated to the LDL ratio. In contrast to the other variables which showed apparently linear interrelationships the H D L ratio gave curved relations to other variables when plotted. DISCUSSION L P lipid composition and L P lipid interrelations were studied in H L P and in an apparently healthy control population of 50-year-old men. It was shown (Table I) that the L P patterns in H L P differ- ed from the normal L P pattern not only in regard of VLDL triglyceride and LDL cholesterol concentra- tion. In spite of great variations in VLDL con- centration the lipid composition, as mirrored by the cholesterol/triglyceride ratio (Table 11) did not differ from normal in types I1 A , I I B and IV. H L P type I11 was characterized by a high cholesterol/ triglyceride ratio (Table 11) caused by a dramati- calIy increased cholesterol level in VLDL (Table 1). The H L P discovered in the investigated popula- tion were moderate as can be expected when a screening of a general population is performed. The mean values for LDL cholesterol in type I1 were about 220 mgl100 ml. Concomitant with the in- creased cholesterol levels, type I1 A and I1 B show- ed increased LDL triglyceride concentrations of a similar relative magnitude resulting in a 'normal' cholesterol/triglyceride ratio (Table 11). Slack and Mills (26) reported a low proportion of triglycerides in the total LDL lipid in patients with familial hyper-P-lipoproteinaemia and suggested the pre- sence of an abnormal L P in this disease. In a recent report (20) type I1 A with tendon xanthomata Upsala .I Med Sci 81 76 B . Vessby et al. were found to have particularly high cholesterol content in relation to triglyceride content. In the mild cases of type I1 without xanthoma tendinosum, most of which probably were of non-familial origin, in the present population, no abnormality was found in regard of the cholesterol/triglyceride ratio. The linear relation between the cholesterol and triglyceride content in LDL in healthy 50-year-old men does not pass through the origin, i.e. the inter- cept ( a ) is significantly different from zero (Table 111). Thus, the ratio cholesterol/triglycerides is a somewhat artificial concept which will increase with decreasing lipid content of LDL. A change of the ratio may not necessarily mean a change in L P composition, especially at low lipid concentrations in LDL. As H L P types I11 and IV (Table 11) showed a low mean ratio in spite of normal LDL cholesterol levels this, however, seems to imply a real change in L P composition in these types of H L P compared to normal L P composition. The density range 1.006-1.063 comprises two LP fractions: LDL, (d= 1.006-1.019 corresponding to Sf 12-20) and LDL, (d=1.01!L1.063, Sf 0-12). The LDL, particles are bigger and relatively more tri- glyceride rich than LDL, (25). Normally LDLz is the quantitatively totally dominating fraction, as is also seen in H L P type I1 (4). The LDL cholesterol/ triglyceride ratio in normals, types I1 A and I1 B is thus mainly reflecting the composition of the LDL, particles. The heterogeneity of the LDL density fraction may explain the lower correlation recorded between cholesterol and triglyceride concentration in LDL than in VLDL (Table 111). Both H L P types 111 and IV were characterized by considerably increased LDL triglyceride concentra- tions when compared with normals (Table I). As the LDL cholesterol levels were normal, the LDL cholesterol/triglyceride ratios in type 111 and IV were low. The lower ratio in type 111 than in type IV was caused by a higher mean value for LDL tn- glycerides (Tables 1-11). The very low LDL ratio in type 111 mirrors a change in the relation between LDL, and LDL, first reported by Gofman et al. (16) in patients with xanthoma tuberosum (e.g. probably type 111). An increased concentration of LDL, in relation to LDL, probably reflects the accumulation of par- ticles metabolically intermediate between VLDL and LDL, because of a block in the normal ca- tabolism of VLDL (21). The group of lipid patterns classified as H L P type IV is probably heterogeneous representing pathogenetically different disorders with an in- crased VLDL concentration in common but with diverging LDL concentration and composition. Some type I V may be genetically related to com- bined hyperlipidaernia (22). Other L P patterns classified as type IV have been suggested to be associated with a retarded conversion of VLDL to L D L because of a relatively decreased VLDL tri- glyceride clearance (28). Also the density fraction > 1.063 contains two or more populations of L P particles with different lipid composition (24). HDL, (d= 1.063-1.125) contains somewhat less triglycerides and more cholesterol than HDL, (d=1.125-1.21) which is also more rich in protein content. Nichols reported (17) that HDL, and HDL, may vary in concentration relatively in- dependent of each other. He did not find any rela- tionship between the cholesterol and triglycerides in HDL in either normals or HLP. In this study low cholesterol levels in H D L in combination with an increased VLDL lipid concentration was seen i type IV which may reflect the redistribution of some L P components (mainly HDL,) from HDL to VLDL after the introduction of triglyceride-rich VLDL into the circulation (12). A reciprocal rela- tionship between HDL and VLDL concentration is present during conditions of varying VLDL con- centration ( I 5 ) . H D L triglyceride concentration was significantly increased in all types of H L P compared with the normal controls (Table I). These ‘HDL triglyceride’ determinations probably measure not only HDL L P triglycerides but also some diglycerides and mono- glycerides produced during the hydrolysis of tri- glyceriderich L P and recovered in the ultracentri- fuge fraction with d> 1.063. The HDL ‘triglyceride’ determinations is certainly both less specific and less accurate ( 2 ) than the triglyceride determination in the other L P classes. One explanation for actually incrased HDL tri- glyceride levels in H L P may be a suggested non- enzymic exchange of HDL esterified cholesterol for VLDL triglycerides secondary to lecithin choles- terol acyl transferase (LCAT) dependent esterifica- tion of free cholesterol in HDL (18). High HDL triglyceride concentrations may possibly be re- garded as reflecting an accelerated break down of triglyceride-rich L P particles with a requirement of disposal of much VLDL surface material. N o signif- icant correlation was found between HDL triglyc- Upsala J Med Sci 81 Lipoprotein composition in 50-year-old men with hyperlipoproteinaemia 77 eride and cholesterol concentrations (Table 111). Thus no conclusion can be drawn from changes in the HDL cholesterol/triglyceride ratio regarding changes in HDL lipid composition. When lipid variables show either a strongly positive or a strongly negative interrelationship, some common determinant of the serum concentra- tions of these lipids seems probable. We have studied linear correlations between L P lipid variables in the random sample of healthy men and in different types of H L P (Table IV-VI). Although the conclusions drawn from this study are valid only for middle-aged men the results were in good agreement with those found in investigations of other populations. The positive relationship between VLDL con- centration and the triglyceride concentration in LDL and HDL as well as the negative relationship between VLDL concentration and HDL choles- terol content is in agreement with earlier studies by Lindgren, Freeman & Nichols (16) and Ewing, Freeman & Lindgren (6). In the present study LDL cholesterol concentra- tion varied remarkably independent of all other L P lipid variables in the controls, types I I A and I I B . In type I V , however, the LDL cholesterol con- centration showed a significant negative relation- ship to VLDL lipid levels and a positive correlation to the cholesterol/triglyceride ratio in VLDL due to the presence of a group of patients with high VLDL triglyceride levels and low LDL cholesterol levels. Gofman & Tandy reported (10) that a regression curve of LDL on VLDL levels studied in two large samples of human males showed a steady rise in LDL levels with increasing levels of VLDL. At very high levels of VLDL, however, the curve sloped downwards and the relationship between those two classes became inverse rather than di- rect. The reason for this may be a retarded inter- conversion of VLDL to LDL because of a relatively deficient triglyceride removal capacity (28). The negative correlation between LDL cholesterol and the VLDL cholesterol/triglyceride ratio is also compatible with this hypothesis. The tendency to a negative relationship between LDL and VLDL ratios seen in H L P type I I A as well as the positive correlation between LDL tri- glycerides and the VLDL ratio in types IIA, IIB and IV may be due to a certain accumulation of particles with a density and composition inter- mediate between VLDL and LDL in some patients associated with a ‘late pre-p’ pattern in d < I .006 on agarose electrophoresis (27). In conclusion all the moderate H L P studied in this investigation showed similar L P lipid interclass interrelationships although they are conventionally divided into different types of H L P because of diverging mean values for L P lipid concentrations. Apart from H L P type I11 and H L P type IV with low LDL cholesterol levels all other types of L P patterns seemed to conform to a common model of L P interconversions where the different H L P types represented different tails of the same spectrum. No clear qualitative differences in L P interrela- tionships between the H L P types were observed. The basically quantitative differences in the statisti- cal correlations between the L P lipids in the healthy controls and types I I A , I I B and some of the type IV may be due to the ‘truncation’ of the spectrum caused by the classification. By simple measures such as dietary modifications, many of these moderately expressed H L P may be transformed to other types of H L P patterns or to a normal L P pattern. H L P type I11 is probably due to a block in the catabolism between VLDL and LDL with ac- cumulation of ‘intermediary’ LP. In type IV a negative correlation exists between VLDL con- centration and LDL cholesterol levels which also at least in some cases, may be caused by a block in the conversion of VLDL to LDL although at an earlier step in the catabolism of triglyceride-rich LP. ACKNOWLEDGEMENTS This work was supported by grants from the Swedish Medical Research Council (19X-204 and 876-4679-01). REFERENCES 1. 2. 3. 4 . 5. Beaumont, J. C., Carlson, L. A . , Cooper, G . R., Fejfar, Z . , Fredrickson, D. S . & Strasser, T . : Classi- fication of hyperlipidemias and hyperlipoprotein- emias. Bull Wld Hlth Org43.891, 1970. Carlson, K. : Lipoprotein fractionation. J Clin Pathol, Suppl. 5: 32, 1973. Carlson, L. A. & Bottiger, L. E.: Ischaemic heart disease in relation to fasting value of plasma tri- glycerides and cholesterol. Lancet I : 865, 1972. Dixon, W. J . (ed.): BMD Biomedical Computer Programs. University of California Press, Berkeley, 1973. Draper, N. R. & Smith, H.: Applied Regression Anal- ysis. John Wiley & Sons Inc., New York, 1966. Upsala J Med Sci 81 78 B. Vessby et al. 6. Ewing, E. M., Freeman, N. K. & Lindgren, F. T.: The analysis of human serum lipoprotein distribu- tion. Advan Lipid Res3: 25, 1965. 7. Fredrickson, D. S. & Lees, R. S . : A system for phenotyping hyperlipoproteinemia. Circulation 32: 321, 1965. 8. Fredrickson, D. S., Levy, R. I. & Lees, R . S . : Fat transport in lipoproteins: an integrated approach to mechanisms and disorders. New Engl J Med 276: 34, 1967. 9. Gofman, J. W., de Lalla, O., Glazier, F., Freeman, N . K., Lindgren, F. T., Nichols, A. V., Strisower, B. & Tamplin, A. R.: The serum lipoprotein transport system in health, metabolic disorders, atherosclerosis and coronary heart disease. Plasma (Milan) 2: 4 13, 1954. 10. Gofman, J. W. & Tandy, R.: In Atherosclerotic Vascular Disease, p. 162. Appleton-Century-Crofts, New York, 1967. 11. Havel, R. J., Eder, H. A. & Bragdon, J. H.: The determination and chemical composition of ultra- centrifugally separated lipoproteins in human serum. J Clin Invest34: 1345, 1955. 12. Havel, R. J., Kane, J. P. & Kashyap, M. L.: Inter- change of apolipoproteins between chylomicrons and high density lipoproteins during alimentary lipernia in man. J Clin Invest 52: 32, 1973. 13. Hedstrand, H. & Vessby, B.: Serum lipid levels and lipoprotein composition in healthy 50-year-old men in Sweden. Upsala J Med Sci 81: 37, 1976. 14. Kannel, W. B., Castelli, W. P., Gordon, T. & McNamara, P. M.: Serum cholesterol, lipoproteins, and the risk of coronary heart disease. Ann Intern Med74: I , 1971. 15. Levy, R. I . , Lees, R. S . & Fredrickson, D. S . : The nature of prebeta (very low density) lipoproteins. J Clin Invest 45: 63, 1966. 16. Lindgren, F. T., Freeman, N. K. & Nichols, A. V.: In Metabolism of Lipids as Related to Athero- sclerosis, p. 62. Ch. C. Thomas, Springfield, Illinois, 1965. 17. Nichols, A . V.: Human lipoproteins and their inter- relationships. Adv Biol Med Phys 1 1 : 109, 1967. 18. Nichols, A. V. & Smith, L.: Effect of very low density lipoproteins on lipid transfer in incubated serum. J Lipid Res 6: 206, 1965. 19. Noble, R. P.: Electrophoretic separation of plasma lipoproteins in agarose gel. J Lipid Res 9: 693, 1968. 20. Olsson, A. G . & Carlson, L. A,: Studies in asympto- matic primary hyperlipidaemia. I. Types of hyperlipo- proteinaemias and serum lipoprotein concentrations, compositions and interrelations. Acta Med Scand, Suppl. 580, 1975. 21. Quarfordt, S. H., Levy, R. I. & Fredrickson, D. S.: The kinetic properties of very low density lipopro- tein triglyceride in type 111 hyperlipoproteinemia. Biochim Biophys Acta296: 572, 1973. 22. Rose, H . G., Kranz, P., Weinstock, M., Juliano, J. & Haft, J. I . : Combined hyperlipoproteinemia. Evi- dence for a new lipoprotein phenotype. Athero- sclerosis20: 51, 1974. 23. Rush, R. L., Leon, L. & Turrel, J.: Automated simultaneous cholesterol and triglyceride determina- tion on the Auto Analyzer@ I1 instrument. Advances in Automated Analysis 1 : 503, 1971. 24. Scanu, A. & Granda, J. L.: Effects of ultracentrifuga- tion on the human serum high-density (1.063 < p < 1.21 g/ml) lipoprotein. Biochemistry5: 446, 1966. 25. Shore, B. & Shore, V.: Some physical and chemical properties of the lipoproteins produced by lipolysis of human serum Sf 2 W O O lipoproteins by postheparin plasma. J Atherosclerosis Res2: 104, 1962. 26. Slack, J. & Mills, G. L.: Anomalous low density lipoproteins in familial hyperbetalipoproteinaemia. Clin Chim Acta29: 15, 1970. 27. Vessby, B . , Carlson, L. A. & Hedstrand, H.: Cholesterol and Triglyceride Composition of Very Low and Low Density Lipoproteins in Hyperlipo- proteinaemias, p. 882. Springer Verlag, Berlin, Hei- delberg, New York, 1974. 28. Vessby, B. & Lithell, H.: Dietary effects on lipo- protein levels in hyperlipoproteinemia. Delineation of two subgroups of endogenous hypertriglycerid- emia. Artery I : 63, 1974. Received January 20, 1976 Address for reprints: B. Vessby, M.D. Department of Geriatrics Box 641 S-75 1 27 Uppsala Sweden Upsala J Med Sci 81