Discordant GH and IGF-1 Results in Treated Acromegaly: Impact of GH Cutoffs and Mean Values Assessment

Context: Discordant growth hormone (GH) and insulin-like growth factor-1 (IGF-1) values are frequent in acromegaly. Objective: To evaluate the impact of different GH cutoffs on discordance rate. To investigate whether the mean of consecutive GH measurements impacts discordance rate when matched to the last available IGF-1 value. Design: Retrospective study. Setting: Referral center for pituitary diseases. Patients: Ninety acromegaly patients with at least 3 consecutive evaluations for GH and IGF-1 using the same assay in the same laboratory (median follow-up 13 years). Interventions: Multimodal treatment of acromegaly. Main Outcome Measures: Single fasting GH (GH(f)) and IGF-1 (IGF-1(f)). Mean of 3 GH measurements (GH(m)), collected during consecutive routine patients' evaluations. Results: At last evaluation GH(f) values were 1.99 2.79 mu g/L and age-adjusted IGF-1(f) was 0.86 +/- 0.44 x upper limit of normality (mean +/- SD). The discordance rate using GH(f) was 52.2% (cutoff 1 mu g/L) and 35.6% (cutoff 2.5 mu g/L) (P = 0.025). High GH discordance was more common for GH(f) <1.0 mu g/L, while high IGF-1 was predominant for GH(f) <2.5 mu g/L (P < 0.0001). Using GH(m) mitigated the impact of GH cutoffs on discordance (GH(m) <1.0 mu g/L: 43.3%; GH(m) <2.5 mu g/L: 38.9%; P = 0.265). At receiver-operator characteristic curve (ROC) analysis, both GH(f) and GH(m) were poor predictors of IGF-1(f) normalization (area under the curve [AUC] = 0.611 and AUC = 0.645, respectively). The prevalence of disease-related comorbidities did not significantly differ between controlled, discordant, and active disease patients. Discussion: GH/IGF-1 discordance strongly depends on GH cutoffs. The use of GH(m) lessen the impact of GH cutoffs. Measurement of fasting GH levels (both GH(f) and GH(m)) is a poor predictor of IGF-1(f) normalization in our cohort.

Automated summary

Discordance between GH and IGF-1 levels in acromegaly patients is influenced by GH cutoff values. The use of GH(m) can mitigate the impact of GH cutoffs and improve predictive accuracy. Fasting GH levels (both GH(f) and GH(m)) are poor predictors of IGF-1 normalization in this cohort.