A paradigm shift in the monitoring of patients with acromegaly: last available growth hormone may overestimate risk

Research output: Contribution to journalArticlepeer-review


CONTEXT: Acromegaly is associated with reduced life expectancy, which has been reported to be normalized if treatment is successful in controlling GH/IGF-I levels.

OBJECTIVE: Most previous studies have invariably used the last available GH/IGF-I, which may be biased as it only assesses exposure at a single point in time. We compared the last available GH/IGF-I analysis to a "time-dependent" and cumulative method, during follow-up to assess risk of mortality in the West Midlands Acromegaly study (n = 501).

RESULTS: Using the last available GH, there was a statistically significant increase in mortality comparing groups as low as GH ≤ 1 μg/L vs >1 μg/L (relative risks [RR] 1.8, P = .03). This was not the case when using the "time-dependent method," where only comparisons of GH values of GH ≤5 μg/L vs >5 μg/L were suggestive of being associated with an increased risk of mortality (RR = 1.5, P = .08). When the time-dependent GH method of analysis was used, the RR of mortality at each level was lower and the associated P value was less significant. Irrespective of using the last available or time-dependent method, when IGF-I was divided into levels according to quartile or arbitrary cutoffs, there was no significant increase in mortality with higher levels.

CONCLUSIONS: This study emphasizes the potential bias of using the latest available GH/IGF-I levels to predict mortality. Our study again highlights the limitations of IGF-I in predicting mortality.


Original languageEnglish
Pages (from-to)478-85
Number of pages8
JournalThe Journal of clinical endocrinology and metabolism
Issue number2
Early online date15 Nov 2013
Publication statusPublished - Feb 2014


  • Acromegaly, Adult, Aged, Combined Modality Therapy, Female, Human Growth Hormone, Humans, Insulin-Like Growth Factor I, Male, Middle Aged, Predictive Value of Tests, Treatment Outcome