Low Testosterone and High C-Reactive Protein Concentrations Predict Low Hematocrit in Type 2 Diabetes

The study was conducted in the Diabetes-Endocrinology Center of Western New York, a tertiary referral center affiliated with the State University of New York and Kaleida Health in Buffalo, New York, and in an endocrinology specialty clinic in Midland, Texas. The study was carried out in 70 male patients (50 from Buffalo and 20 from Midland) referred to the centers for management of type 2 diabetes. Patients with a known history of primary or secondary hypogonadism, hypopituitarism, renal failure, cirrhosis, glucocorticoid therapy, or known HIV infection were excluded from the study. Demographic parameters were collected, and height, weight, glucose, and HbA_1c (A1C) were measured. The age of patients (means ± SE) was 56.3 ± 1.6 years (range 24–78), weight was 103.2 ± 2.9 kg (49.5–156), BMI was 32.8 ± 0.9 kg/m² (18.4–54.1), and A1C was 7.74 ± 0.25% (4.7–14). The clinical and biochemical features of the patients are summarized in Table 1. Fasting blood samples were obtained to measure plasma hemoglobin, hematocrit, serum total testosterone, sex hormone–binding globulin (SHBG), luteinizing hormone (LH), follicle-stimulating hormone (FSH), prolactin, glucose, A1C, and plasma CRP. Serum total iron, iron-binding capacity, ferritin, vitamin B₁₂, and folate concentrations were measured in all 16 patients who had anemia.

Total testosterone was measured by a solid-phase radioimmunoassay (Coat-A-Count; Diagnostic Products, Los Angeles, CA). The lower limit of normal for total testosterone in our clinical laboratory is 10.4 nmol/l (300 ng/dl). SHBG was tested at Specialty Laboratories (Santa Monica, CA) by an immunochemiluminometric assay. Anemia was defined as hemoglobin <13 g/dl or hematocrit <39.0% as per the World Health Organization definition.

Calculated free testosterone (cFT) was determined from SHBG and total testosterone using the method of Vermeulen et al. (8), using a computer program and Web site address (http://www.issam.ch/freetesto.htm) provided by Dr. T. Fiers, University Hospital Ghent, Ghent, Belgium. This cFT value has been shown to correlate very well (r = 0.9) with free testosterone measured by equilibrium dialysis (9). For cFT, 0.225 nmol/l (6.5 ng/dl) was taken as the lower limit of normal (10). It is known that cFT values are generally 10–15% higher than those for free testosterone measured by equilibrium dialysis. Bioavailable testosterone (non–SHBG-bound testosterone) was also calculated similarly using SHBG and testosterone. The lower limit of normal was considered to be 5.2 nmol/l (150 ng/dl) (10). LH and FSH were measured by chemiluminescent immunometric assays. The lower limit of total testosterone was 300 ng/dl, that for cFT was 6.5 ng/dl, and that for bioavailable testosterone was 150 ng/dl in normal subjects. Hypogonadism was defined as cFT <6.5 ng/dl. Hypogonadotrophic state was defined as inappropriately low FSH and LH concentrations for the concomitant subnormal cFT concentrations. Thus, LH and FSH concentrations in the normal reference range would be considered to be “low” when associated with low cFT concentrations. CRP levels >3 mg/l were termed “high” because those concentrations are known to be associated with increased cardiovascular events. Plasma CRP concentrations were measured using a high-sensitivity enzyme-linked immunosorbent assay kit from Alpha Diagnostics International (San Antonio, TX).

Erythropoietin was measured in duplicate using the Advantage erythropoietin chemiluminescence immunoassay (Nichols Institute Diagnostic, San Clemente, CA), which has a sensitivity of 1.2 mU/ml and a coefficient of variation <6%. The mean of the two assays was used in the analysis. Glomerular filtration rate (GFR) was estimated by calculating creatinine clearance from the serum creatinine concentration using the Cockcroft-Gault formula.

Data are presented as means ± SE. Student’s t test was used to compare parametric data, and the Mann-Whitney rank-sum test was used to compare nonparametric data. Fisher’s exact test or the χ² test was also used to compare the groups whenever appropriate. Spearman correlation (for nonparametric data) and Pearson correlation (for parametric data) were used to establish correlations. Multiple regression analysis between variables was performed if there was more than one independent variable. P < 0.05 was considered significant. Sigma Stat software was used for analysis.

The plasma total testosterone and cFT concentrations (means ± SE) in hypogonadal men (n = 37) were 212.84 ± 13.56 ng/dl (range 20–291) and 4.43 ± 0.22 ng/dl (0.26–6.39), respectively, and were significantly lower (P < 0.01) than those in eugonadal men (n = 33; 469.35 ± 32.39 ng/dl [305–1,428] and 9.19 ± 0.38 ng/dl [6.56–13.9], respectively). The mean CRP concentrations were 6.5 mg/l in hypogonadal patients and 3.2 ng/dl in eugonadal patients (P < 0.001); 40 patients (57.1%) had CRP ≤3 mg/l and 30 (42.9%) had CRP >3 mg/l. Table 2 summarizes the characteristics of patients with low and high CRP concentrations. Mean hematocrit was 40.6 ± 1.1% in hypogonadal men, whereas it was 43.3 ± 0.7% in eugonadal men (P = 0.011) (Fig. 1). The hematocrit was 42.7 ± 0.7% in patients with CRP <3 mg/l compared with 39.9 ± 1.1% in patients with CRP >3 mg/l (P = 0.05) (Fig. 2). Anemia was observed in 16 of 70 (23%) patients. In 15 of these 16 anemic patients, the anemia was normocytic and normochromic and was not associated with iron, folate, or vitamin B₁₂ deficiency. In one patient, it was microcytic and hypochromic and was associated with iron deficiency. Fourteen of 15 (93%) patients with normocytic normochromic anemia had low cFT concentrations. Erythropoietin concentrations were normal or elevated in all of the 11 patients with normocytic normochromic anemia in whom they were measured. Bone marrow biopsy was available in one patient with hypogonadism and normocytic normochromic anemia. It demonstrated normal iron stores and had normal cellularity (40%) with normal erythroid, myeloid, and megakaryocytic cell lines. The prevalence of normocytic normochromic anemia was 37.83% in hypogonadal men, whereas it was 3% in eugonadal men (P < 0.05). The prevalence of anemia was 13.2% in patients with CRP <3 mg/l, whereas it was 33.3% in those with CRP >3 mg/l (P < 0.05). The frequency of anemia in those with both low cFT and high CRP was 50%. SHBG concentrations tended to be lower in patients with anemia and in those with elevated CRP.

Hematocrit was related to total testosterone (r = 0.36; P < 0.001) and cFT (r = 0.46; P < 0.0001) (Fig. 3). There was an inverse relationship between plasma CRP concentrations (r = 0.41; P < 0.0004) and hematocrit (Fig. 4). There was also an inverse relationship between plasma cFT and CRP concentrations (r = −0.27; P = 0.02) (Fig. 5). Serum creatinine concentrations were not significantly different between hypogonadotrophic hypogonadism and eugonadal patients, nor was there a relationship between serum creatinine clearance and hematocrit. GFR was estimated from serum creatinine concentration of the patients using the Cockcroft-Gault formula. The mean calculated creatinine clearance was 130 ± 8.6 ml/min in the hypogonadal group, whereas it was 135 ± 6.1 ml/min in the eugonadal groups, and the difference was not significant (P = 0.33). No significant relationship was found between GFR and hematocrit (r = 0.2; P = 0.8). On multivariate analysis, both cFT (P < 0.001) and CRP (P < 0.01) but not BMI were related independently to hematocrit. In an another multivariate analysis using CRP as the dependent variable and cFT, BMI, and age as independent variables, only cFT and age were independently related to CRP (P < 0.001 and 0.006, respectively). SHBG concentrations were inversely related to CRP (r = −0.23; P < 0.02) but were not related to either the hematocrit or cFT.

This demonstration of a significantly lower hematocrit in hypogonadal men and a direct relationship between the hematocrit and testosterone in type 2 diabetic subjects, described for the first time, suggests that a low testosterone concentration may contribute to the pathogenesis of the mild anemia in these patients. Testosterone is known to stimulate erythropoiesis in the bone marrow and to increase the hematocrit (3). Consistent with a diminished erythropoietic drive was the fact that in 15 of 16 anemic patients, the anemia was normocytic normochromic and 14 of 15 patients had a subnormal cFT. Only 1 patient of 16 had iron deficiency with a hypochromic microcytic anemia. None of the 15 patients with normocytic normochromic anemia had iron, folate, or vitamin B₁₂ deficiency. Eleven of 15 patients with normocytic normochromic anemia had erythropoietin concentrations measured; all 11 had high normal or elevated erythropoietin concentrations. Thus, the anemia was not due to erythropoietin deficiency.

The highly significant inverse relationship between CRP concentrations and the hematocrit, independent of testosterone concentrations, suggests that inflammatory processes in type 2 diabetes probably also suppress the hematocrit. CRP has for a long time been considered a marker of systemic inflammation; therefore, it is also a prognosticator of cardiovascular events (11–13) because atherosclerosis is a chronic inflammation of the arterial wall. However, there is now evidence that CRP may be a mediator of inflammation (14–16). It induces intercellular adhesion molecule-1 and macrophage inflammatory protein-1 in endothelial cells in vitro and probably exerts this effect through the FcII receptor (17). More recently, it has been shown to exert a proinflammatory effect after an injection into normal human subjects in vivo (18). Thus, the relationship of CRP with hematocrit may either be through the direct action of CRP or through other inflammatory mediators associated with CRP or both. Inflammatory mechanisms may affect the hematocrit in two ways. First, they may suppress erythropoietin secretion (4), and second, they may cause increased apoptotic death of red cell precursors, resulting in no increase in erythropoiesis despite elevated erythropoietin levels (4,19–21). These mechanisms are relevant to the pathogenesis of anemia of chronic inflammatory disease. Thus, the mild anemia of type 2 diabetes may in part be attributable to processes similar to those involved in chronic inflammatory disease in addition to the contribution by low testosterone concentrations. The overwhelming predominance of the normocytic normochromic picture in the anemic patients is also consistent with marrow suppression secondary to inflammatory mechanisms. Elevated erythropoietin levels associated with anemia in these patients suggest an inadequate response to erythropoietin rather than a deficiency of erythropoietin.

The relationship of inflammatory mechanisms with hematocrit becomes even more intricate because these mechanisms may be involved in the pathogenesis of hypogonadotrophic hypogonadism itself in patients with type 2 diabetes. We have previously reported and confirmed again in this study that there is a highly significant inverse relationship between plasma testosterone and CRP concentrations (V.B., R.T., S.D., A. Chandel, A.C., H.G., P.D., unpublished observations). It has previously been shown that insulin may facilitate gonadotropin-releasing hormone release from hypothalamic neurons in vitro and that interference with insulin signal transduction may reduce gonadotropin-releasing hormone secretion (22). Because inflammatory mediators interfere with insulin signal transduction, they may contribute to the pathogenesis of hypogonadotrophic hypogonadism. Consistent with our previous report, LH and FSH concentrations in patients with hypogonadotrophic hypogonadism were significantly lower than those in eugonadal patients.

One recent European study (23) showed a mild elevation of hematocrit in patients with type 2 diabetes. The lowest quartile of hematocrit in this study had values that were similar to those in patients with hypogonadotrophic hypogonadism in our study. Although we cannot confirm elevated hematocrit in type 2 diabetes in our study, it would be worth investigating whether low hematocrit in other studies is associated with hypogonadotrophic hypogonadism.

We conclude that the hematocrit is significantly lower in hypogonadal men with type 2 diabetes compared with eugonadal men and that there is a significant direct relationship between cFT and hematocrit. There is also an inverse relationship between hematocrit and CRP. Thus, both a low testosterone concentration and inflammatory mechanisms may play an important role in the pathogenesis of the low-grade anemia observed in patients with type 2 diabetes. Because high CRP concentrations are known to be associated with atherosclerosis and there are early data showing that low testosterone concentrations may be associated with increased cardiovascular events (24), the anemic diabetic patient may have an increased risk of atherogenesis.