The predictors of recovery from diabetes mellitus following neurosurgical treatment of acromegaly: A prospective study over a decade
Keywords: Acromegaly, diabetes mellitus, growth hormone, insulin-like growth factor-1
Although the unidirectional relation of acromegaly being a cause of diabetes mellitus (DM) is beyond doubt, diabetes influences the diagnosis and evaluation of acromegaly. Concomitant diabetes influences the diagnosis and evaluation of acromegaly. In one of the early reports by Davidoff and Cushing, the incidence of DM in their 100 patients with acromegaly was 12%. This figure was still higher (17%) in the January 1940 bulletin of “The association for the study of internal secretions” by Kunstadter. Since then, multiple registries from all over the word have reported an increasing prevalence of DM among patients with acromegaly. Diabetic ketoacidosis (DKA) may be one of the rare presenting manifestations of acromegaly in some cases.
Excess amounts of growth hormone (GH) and insulin-like growth factor (IGF)-1 are associated with hepatic and peripheral insulin resistance. This and post-receptor signaling mechanisms also contribute to the development of DM. Hyperinsulinemia, insulin resistance, and diabetes are well-recognized cardiovascular risk factors in the general population and may contribute as well to the increased cardiac morbidity and mortality of acromegalic patients. In a previous study, age, body mass index (BMI), and hypertension were determined to be the significant risk factors for the development of DM in patients with acromegaly, like that seen in the general population. There is only a single study with a small number of patients from our country addressing these issues. Therefore, we conducted this study evaluating the prevalence, predictors, and the natural course of DM in 151 patients with acromegaly treated at a single tertiary care center.
This study included 151 patients with acromegaly treated at the Departments of Endocrinology and Neurosurgery, Post Graduate Institute of Medical Education and Research, Chandigarh, India, in the past 10 years (starting from June 2007). The inclusion criteria for the study were the same as mentioned in our previous series, which were as follows: (a) nadir GH levels during an oral glucose tolerance test (OGTT) >1 ng/mL, (b) IGF-1 levels above the normal range matched for age, and (c) clinical and radiologic features consistent with the diagnosis of acromegaly associated with a pituitary adenoma. The diagnosis of DM and pre-diabetes was made as per the American Diabetes Association criteria.
In the presence of grossly uncontrolled diabetes, the inclusion criteria were any two of the following: (a) random GH levels during an OGTT (with uncontrolled blood glucose) >5 ng/mL and IGF-1 levels more than the age-matched normal range; or, (b) nadir GH levels during an OGTT (performed 2 weeks after glycemic control) >1 ng/mL and IGF-1 levels more than the age-matched normal range; and, (c) insulin-like growth factor-binding protein-3 (IGFBP-3) raised above the normal range.
At the initial visit, an elaborate history-taking and clinical examination were done along with assessment of the baseline anterior pituitary hormonal profile, including HbA1c, GH, and IGF-1 levels. Besides the routine investigations, fasting insulin and fasting C-peptide (2012 onward) were also measured. GH levels were measured by electrochemiluminescence (IMMULITE, Siemens, Germany) until 2011, followed by Cobas 6000 (Roche–Hitachi, Basel, Switzerland). IGF-1 was measured by chemiluminescence (LIAISON, DiaSorin, Italy) throughout the study period. The World Health Organization (WHO) standard 98/574 was used for GH assay and 02/254 for IGF-1 assay. HbA1c was measured by high-performance liquid chromatography (BioRad variant II, Hercules, CA, USA). Fasting insulin and fasting C-peptide were measured by chemiluminescence (Cobas 6000; Roche–Hitachi). Glucose-suppressed GH was done in all patients, as per the protocol.
Contrast-enhanced magnetic resonance imaging (MRI) of the hypothalamo-pituitary axis was done in all patients preoperatively and during the follow-up visit, to look for any residual tumor. Similarly, GH and IGF-1 levels were measured at 12 weeks post-operatively. HbA1c, fasting insulin, and fasting C-peptide were also measured during the follow-up visit. All biopsy specimens were sent for histo-pathological examination and for immunostaining of anterior pituitary hormones [GH, luteinizing hormone (LH), follicle stimulating hormone (FSH), thyroid stimulating hormone (TSH), alpha melanocyte-stimulating hormone (ACTH), and prolactin (PRL)). Cure was defined as follows: (a) nadir GH levels during an OGTT <0.4 ng/mL or random GH <1 ng/mL; and (b) IGF-1 levels within the normal range matched for age. The cure of diabetes was defined as HbA1c ≤6.5% without any drugs; and, partial improvement of DM was defined as a decrease in insulin requirement of more than 30%, conversion from insulin to oral hypoglycemic agent (OHA), or more than 50% decrease in OHA requirement.
Statistical Package for the Social Sciences (SPSS) 21 software (IBM Corp., Armonk, NY, USA) was used for statistical analyses. Proportions were compared using a chi-square or Fisher's exact test, wherever appropriate. After evaluating for normality of data, non-parametric tests were chosen. Univariate analyses of continuous variables used Mann Whitney U-test across binary categories and Kruskal-Wallis test across multiple categories. The bivariate relationship between two continuous variables was assessed using Spearman's rank-order correlation coefficient. Multivariate analysis was conducted using binary logistic regression for evaluating the recovery from diabetes, with mandatory significance of the model coefficient being <0.05, after adjusting for other known prognostic factors such as age, BMI, pre-operative HbA1c, and post-operative IGF-1.
There were 80 (53%) male and 71 (47%) female patients in our study group with their median age being 36 years (range: 6–70 years). The median BMI was 27.1 kg/m 2 (range: 18.9–39.1 kg/m 2) and the median duration of acromegaly was 4 years (maximum duration of 35 years). Thirteen of our patients were diagnosed incidentally when they were visiting various departments of our institute for accompanying comorbidities such as obstructive sleep apnea, arthritis, and tonsillomegaly. The median GH level in our subjects was 25 ng/mL (range: 0.60–1290 ng/mL), and for IGF-1, the median level was 811 ng/mL (range: 111–9111 ng/mL).
The median duration of DM in our patients was 2 years (range: 2 months–20 years) and 53 patients were diagnosed with DM at the time of their initial visit. Overall, four patients presented with DKA as the initial manifestation, in the absence of obvious precipitating factors for DKA. In all, 93 (61.6%) subjects were diabetic and 20 (13.2%) suffered from pre-diabetes (impaired fasting glucose or impaired glucose tolerance). The age-wise distribution of patients with diabetes is shown in [Figure 1]. Acanthosis nigricans was present in 61% patients and skin tags were present in 45.7% patients. The median HbA1c was 6.4% (range: 4.3%–18%). Approximately 12.6% patients had a family history of DM in first-degree relatives. Overall, six patients died during the follow-up period, with one succumbing to DKA (had refractory seizures and aspiration pneumonia at the terminal stage). [Table 1] shows the comparison between diabetic and non-diabetic populations, in terms of baseline characteristics. Both populations were comparable, except that those patients suffering from diabetes were significantly older.
Pre-operatively, 19.4% of the diabetic patients were receiving only oral drugs, 11.8% were taking only insulin therapy, and 29% of patients were receiving insulin and oral drugs. The majority [139 (92.1%)] had a macroadenoma, 5 (3.3%) had a microadenoma, and 7 (4.6%) had a mesoadenoma. The other comorbidities, such as hypertension, were present in 62 (41.1%) patients; and, dyslipidemia was present in 37.7% patients, with an elevation of low-density lipoproteins (LDLs) being the most common manifestation, followed by hypertriglyceridemia. Cardiomyopathy was present in only two patients. Six patients had concomitant malignancies in the form of carcinoma ovary, breast, colon, chronic myeloid leukemia (CML), multiple myeloma, and follicular lymphoma. Before surgery, 5.3% patients received somatostatin therapy to control the grossly deranged glycemic profile. Three patients in our study had diabetic retinopathy, including one patient with proliferative retinopathy with clinically significant macular edema (CSME), for which he received laser therapy and injection bevacizumab [Figure 2]. All these patients had diabetes duration of more than 5 years. The majority of the patients underwent surgery for pituitary adenoma in the form of either trans-sphenoidal surgery or a craniotomy, except seven patients in whom surgery was not performed; and, one patient each received gamma knife and external beam radiotherapy as the primary mode of treatment. Pre-operatively, fasting plasma insulin [median (range): 13.99 μU/mL (0.51–88.4 μU/mL)] and C-peptide [median (range): 3.71 ng/mL (0.08–22 μU/mL)] were available for 53 patients, but post-operative values could be measured only in 25 patients. There was no significant difference between the pre-operative and post-operative values of both the parameters in either group.
Post-operatively, 31.1% patients were cured of acromegaly, as per the criteria mentioned previously. In terms of tumor size, 60% (3 patients) with a microadenoma, 57% (4 patients) with a mesoadenoma, and only 28.7% (37 patients) harboring a macroadenoma achieved cure. The overall recovery from diabetes was 46.8% (42.6% had complete recovery, 4.2% had partial recovery). On the contrary, 77.8% patients recovered from pre-diabetes. Following surgical treatment, the median HbA1c decreased significantly from 6.4 to 5.5 (P < 0.001). Immunohistochemistry showed that 55.6% adenomas were positive only for GH, 22.5% for GH and PRL, 3.3% did not stain for any hormone, and the data were unavailable for 24 patients (21.9%).
Among the 113 patients with acromegaly with DM and pre-diabetes, 77 had complete diabetic outcome data at follow-up. About 19 of 26 (73.1%) patients with biochemical remission had recovery from diabetes when compared with 14 of 42 (33.3%) patients without biochemical remission (P = 0.001). The patients who recovered from DM had significantly lower IGF-1 (P < 0.001) and GH (P = 0.02) levels compared with others [Figure 3]. The change in HbA1c had significant rank-order correlation with change in GH levels following surgery [Figure 4] (P = 0.02).
Many patients showed discordance between the GH and IGF-1 levels in the post-operative period, with 13.33% (20 patients) having a high GH but a low IGF-1 level and 3.33% (5 patients) having a low GH but a high IGF-1 level. Recovery from diabetes had a greater association with IGF-1 than GH, especially among those with discordant GH and IGF-1 levels [60% (12/20) in normal IGF-1 and high GH vs. 20% (1/5) in high IGF-1 and normal GH] post-operatively. Moreover, 65.5% of those with radiological remission had recovery from diabetes compared with 34.1% of those with radiological evidence of residul tumor (P = 0.01).
Multivariate analysis was performed using binary logistic regression, with recovery from diabetes as the dependent variable and age, pre-operative HbA1c, post-operative IGF-1 levels, and BMI as independent variables. The analysis showed that only post-operative IGF-1 levels determined recovery from DM (P = 0.01) independently. There was an insignificant decline in fasting insulin and C-peptide post-operatively, compared with the pre-operative levels.
Among the manifestations of pituitary adenomas, acromegaly poses a medico-surgical challenge due to the associated metabolic comorbidities, leading to an overall greater clinical burden.,,
The prevalence of DM and pre-diabetes in our study was 61.6% and 13.2%, respectively. A family history of DM was present in 12.6% patients. Post-operative IGF-1 and a visible residual tumor were the only major determinants of recovery from DM.
There has been a consistent trend of the rising prevalence of DM in patients with acromegaly since the article written in the early 1940 by Kunstadter. The authors reported a 17% prevalence rate in their 153 patients with acromegaly. A family history of DM was present in 21% of their patients with diabetes, whereas the same figure in this study was 12.6%. The most common presenting manifestations of diabetes in our study were the classic osmotic symptoms, weight loss, and rarely DKA. The prevalence rate of diabetic retinopathy was only 3.3% in our study when compared with 2.2%–12.5% in previously published articles., However, in our study, we reported only patients with manifested retinopathy who had undergone a detailed fundoscopic examination, including fluorescence angiography.
The predominant mechanism underlying DM in acromegaly is peripheral and hepatic resistance towards insulin action, caused by increased GH levels. The increased levels of free fatty acids (FFAs) produced by increased lipolysis (due to the action of GH) reduce the use of glucose as fuel in the muscles by reducing glucose uptake. Moreover, GH also promotes gluconeogenesis and inhibits muscle glycogen synthase activity. At the molecular level, insulin receptor substrate 1 (IRS 1) and phosphatidylinositol 3-kinase (PI3K) are responsible for post-receptor resistance toward insulin action, as the density of insulin receptors remains unaltered in acromegaly.,,,
Diabetes may alter the diagnostic criteria of acromegaly, especially if the glycemic profile is grossly deranged. This is exemplified by the study of Alexander and Irina, who measured GH and IGF-1 levels in patients with acromegaly with and without diabetes and revealed that both GH and IGF-1 levels were lower in patients with acromegaly suffering from diabetes. Moreover, our own study showed that OGTT was discriminatory in the diagnosis of acromegaly even in the presence of poor glycemic control. This article also highlights that IGFBP-3 levels are not influenced by the degree of glycemic control, although IGF-1 levels should be dealt with caution in poorly controlled DM.
As shown in [Table 1], the patients suffering from diabetes were 6 years older than the non-diabetic population, which was comparable to the Belgian acromegaly registry and the study from Moscow by Dreval et al.,,, On the contrary, only 5% patients in this registry were on both oral drugs and insulin, whereas 29% of our patients were receiving both therapies. The prevalence of hypertension and dyslipidemia was not different between the diabetic and non-diabetic patients in our study, unlike the study from Brazil where hypertension was more prevalent in the diabetic population (73% vs.46%, P = 0.04). Similarly, the BMI was comparable in both populations in our study, which was different from the French Acromegaly Registry that showed diabetic patients had a significantly higher BMI when compared with non-diabetic patients with acromegaly.
Ironically, the IGF-1 levels and the duration of acromegaly were less in diabetic patients with acromegaly compared with non-diabetic patients in our study, a finding that was completely in disagreement to the study by Dreval et al., and the Mexican Acromegaly Registry., However, GH levels were higher in patients with acromegaly with diabetes when compared with the non-diabetic population, which was in line with the Mexican Registry and the classic study by Nabarro., Multiple logistic regression analysis performed in our study revealed that only post-operative IGF-1 levels were predictive of DM cure, which was like the findings of multiple regression analysis done in the Brazilian acromegaly study.
The main strength of our study was the elaborate follow-up conducted to determine the outcome of DM in patients with acromegaly, post-operatively. Previous registries have highlighted the predictors, prevalence, and determinants of cure, but the outcome of diabetes in terms of recovery, persistence, improvement, and worsening has never been stressed upon. In our study, the presence of residual lesion also emerged as a significant determinant of persistence of DM. Post-operatively, 23.8% patients completely recovered from diabetes and 17.9% showed improvement in glycemic control (on lesser drugs when compared with the pre-operative medications). A comparison of different acromegaly registries and this study is depicted in [Table 2].
DKA, as a presenting manifestation of acromegaly, is rare. There are isolated case reports or small case series showing a prevalence of 1%. In this study, all the patients presenting with DKA were males in their third decade of life, except one who was 48 years old, and the duration of diabetes varied from 3 months to 3 years. The high fasting plasma insulin and C-peptide levels could be due to the compensatory beta cell response to overcome the high peripheral insulin resistance, which came down following surgery, similar to the observations made in a previous study.
One of the limitations of our study was that the data were incomplete for a few patients, especially in terms of the fasting insulin and the fasting C-peptide, levels, and the lag period between the diagnosis of acromegaly and the development of DM; hence, we did not analyze the same in our study. Similarly, we did not have markers such as glutamic acid decarboxylase (GAD)-65 to exclude auto-immune DM, especially in those with DKA. Studies are also needed to evaluate whether the superior visualization provided by the endoscopic approach over the microscopic approach offers any benefit on metabolic control of acromegaly.,
The evolving role of newer medical modalities for treating pituitary adenomas needs to be further studied for their impact on glycemic status., In conclusion, we reiterate that prevalence of DM among patients with acromegaly is on the rise. Nearly half of DM or pre-diabetes patients with acromegaly had glycemic recovery, influenced by biochemical and radiological remission. Post-operative IGF-1 appears to be the strongest independent determinant of diabetic recovery.
We are deeply saddened due to the tragic demise of Prof. K. K. Mukherjee who was our dear friend, senior colleague, and mentor. He was instrumental in encouraging us to write this article before his demise.
Financial support and sponsorship
Conflicts of interest
There are no conflicts of interest.
[Figure 1], [Figure 2], [Figure 3], [Figure 4]
[Table 1], [Table 2]