Subjects and Methods
After IRB approval, individuals 18–90 years of age with a history of acromegaly were recruited from dedicated study sites, including the Massachusetts General Hospital Neuroendocrine and Pituitary Tumor Clinical Center, the Brigham and Women's Hospital Pituitary and Neuroendocrine Program and the Swedish Pituitary Center, from April 2013 to December 2017. Subjects were additionally recruited from national advertising efforts during this period for off-site visits. Subjects were included if they belonged to one of the following three groups: (a) uncontrolled acromegaly with an IGF-1 level above the normal range studied prior to initiation of any medical therapy; (b) controlled acromegaly receiving somatostatin analog monotherapy (octreotide acetate depot or lanreotide); or (c) controlled acromegaly receiving GH receptor antagonist monotherapy (pegvisomant). Subjects receiving both medications concomitantly were excluded from participation. Hypothyroidism, adrenal insufficiency and hypogonadism were documented by physician review of medical history and active medication list. Subjects with untreated hypothyroidism or adrenal insufficiency were excluded. Daily replacement glucocorticoid dose for patients with adrenal insufficiency was determined by chart review and converted to prednisone equivalents based on standard conversion tables (hydrocortisone dose/4 = prednisone equivalent). Additionally, recent initiation or discontinuation of gonadal steroid therapy as well as pregnancy or nursing were all exclusionary.
A total of 124 subjects were studied: n = 60 receiving somatostatin analogs (SSA group); n = 31 receiving pegvisomant (PEG group); n = 33 with active acromegaly on no medical therapy (ACTIVE group). After IGF-1 levels were run on batched samples, seven subjects were excluded from analysis for IGF-1 level greater than the upper limit of normal (n = 5 SSA, n = 2 PEG) and one subject was excluded from analysis in the ACTIVE group for an IGF-1 level within the normal range.
Within the SSA group, 40% were receiving octreotide acetate depot and 60% were receiving lanreotide. The mean monthly octreotide acetate equivalent dose was 25 ± 10 mg (range 1.64–40 mg monthly) based on a potency of 1:3 (octreotide acetate depot to lanreotide). Pegvisomant was self-administered anywhere from 2 to 7 times per week; the calculated mean daily pegvisomant dose was 12.6 ± 6.9 mg (range 2.8–30 mg daily). Twenty-four (83%) of the PEG subjects had previously been on a somatostatin analog; mean duration of somatostatin analog use prior to initiating pegvisomant was 47.6 months (range 3–156 months). Reasons for discontinuation of prior somatostatin analog therapy were lack of efficacy (n = 15, 52%), hyperglycaemia (n = 3, 13%), GI side effects (n = 10, 42%) and lack of insurance coverage (n = 1, 4%), with four subjects experiencing more than one reason for discontinuation. Specific GI side effects prompting SSA discontinuation in the ten subjects included diarrhoea (n = 3); bloating, abdominal cramping and diarrhoea (n = 2); constipation (n = 1); abdominal pain (n = 1); pale stool (n = 1) and unspecified GI symptoms (n = 2).
Acromegaly treatment history, past medical history, hormonal deficiencies and hormone replacement therapies were identified through patient interview and review of primary endocrine provider notes and medication lists. Presence of diabetes mellitus was determined by patient interview, chart review and use of medications used to treat diabetes mellitus.
QoL was assessed using the following self-rated scales (* in list below indicate the four primary end-points): Acromegaly QoL Questionnaire (AcroQoL) (n = 5 domains, including AcroQoL Global Score* as well as Physical, Psychological, Appearance and Personal Relationship Domains), Rand 36-Item Short Form Survey (SF-36) (n = 10 domains, including Physical Component Summary Score*, Mental Health Component Summary Score* as well as Physical Function, Role Limitations Due to Physical Health, Role Limitations Due to Emotional Problems, Energy/Fatigue, Emotional Well-being, Social Function, Pain and General Health Domains) and Gastrointestinal QoL Index (GIQLI) (n = 6 domains, including GIQLI Global Score* as well as Physical Function, Social Function, Emotions, Gastrointestinal Symptoms and Medical Treatment Domains). Surveys were administered via phone, paper or online REDCap form per subject preference. For all questionnaires, higher scores indicated better QoL.
Serum IGF-1 levels were measured by LC/MS-MS (Quest Diagnostics). Total and free thyroxine (FT4) levels were measured by immunoassay (Mayo Laboratories). Total testosterone was measured by LC/MS-MS and percent free testosterone by equilibrium dialysis (Mayo Laboratories). HbA1c was obtained from all available samples (Quest Diagnostics). Fasting blood samples were obtained in a subset of the subjects without diabetes mellitus (n = 67). Of those subjects, 31 also underwent formal oral glucose tolerance testing (OGTT). Insulin was measured by immunoassay and glucose by spectrophotometry (Quest Diagnostics). HOMA-IR was calculated as (fasting insulin × fasting glucose)/405. Area under the curve (AUC) values for glucose and insulin were calculated using the following equation ('level' refers to plasma glucose [glucose AUC] or insulin level [insulin AUC]): (15 × 0 minute level) + (30 × 30 minute level) + (30 × 60 minute level) + (30 × 90 minute level) + (15 × 120 minute level).
Continuous variables are reported as mean ± standard deviation and categorical variables as n (%). Variables were log-transformed prior to statistical analysis. The two-sample t test was used for two-way comparisons and Fisher's least significance test was used to control for multiple comparisons in 3-group analyses. Pearson's correlation coefficients were performed for univariate correlations. Multivariable regression models were also constructed for analysis of determinants of QoL. Statistical significance was set as two-tailed P ≤ .05. All statistical analyses were performed using JMP Pro Statistical Database Software (Version 12; SAS Institute). The study was powered a priori for the four primary end-points (AcroQoL Global Score, SF-36 Physical Component Summary Score, SF-36 Mental Health Summary Score and GIQLI Global Score). For the comparison of PEG vs ACTIVE, with 30 patients per arm we calculated that we would be able to detect a difference of 0.89 times the standard deviation of the variables at a P-value of .0125 (0.05 divided by 4 primary end-points) and a difference of 0.74 times the standard deviation at a P-value of .05. For other comparisons, we could detect a difference of 0.76 with Bonferroni correction and 0.64 without.
Clin Endocrinol. 2021;94(1):58-65. © 2021 Blackwell Publishing