Growth Hormone Replacement Therapy Is Not Associated with Any Increase in Mortality

Bengt-Åke Bengtsson

Research Centre for Endocrinology and Metabolism Sahlgrenska University Hospital Goteborg S-413 25, Sweden.

H. P. F. Koppeschaar

Acadenic Hospital Utrecht NL-3584, the Netherlands

Roger Abs

University Hospital Antwerp B-2650 Egeden, Belgium

Helge Bennmarker, Elisabeth Hernberg-Ståhl, Bjorn Westberg and Patrick Wilton

Pharmacia and Upjohn Outcomes Research Stockholm SE-19287, Sweden

John P. Monson

St. Bartholomew’s Hospital London EC 1A7BE, UK

U. Feldt-Rasmussen

Rigshospitalet Copenhagen DK 2100, Denmark

Christian Wüster, on behalf of the KIMS Study Group

University Hospital Heidelberg D-6912, Germany

Recently, Takala et al. (1) reported the outcome of GH treatment of severely ill patients necessitating intensive care unit (ICU) therapy in two placebo-controlled multicenter trials, each comprising 250 patients. The primary end point was length of stay in the ICU. Patients were treated with high doses of GH either 16 or 24 IU per day, depending on body weight. The outcome showed that mortality was significantly higher in the GH-treated group (42%) compared to placebo (18%). The cause(s) of this unfortunate outcome remains obscure (1).

It has also been previously demonstrated that hypopituitary adults receiving conventional replacement therapy but not GH have an increased mortality. Rosén et al. (2) reported a significantly increased mortality with a standard mortality rate (SMR) of 1.8. This result was confirmed in a later study by Bülow et al. (3), who reported a SMR of 2.17. In addition, Bates et al. (4) reported a SMR of 1.73. The consequences of GH deficiency (GHD) in adults and the effect of replacement therapy have been delineated during the last 10 yr (5), and in many countries the regulatory authorities have approved adult GH replacement therapy. In light of the unfortunate outcome of the ICU trials, we have performed an analysis of mortality during replacement treatment with GH to deficient adults using the largest database available for these patients. KIMS, the Pharmacia and Upjohn International Metabolic DATABASE, is a long-term open pharmacoepidemiological survey that was initiated in 1994. Adult hypopituitary patients with GHD are treated with GH in a conventional clinical setting. The cohort is followed for safety and efficacy of treatment. When patients enter the survey, a baseline case record form is completed and patients are then followed prospectively at least annually. Initially, when adjusting doses, patients are followed more frequently. Patients previously treated, on treatment or not previously treated with GH may be included in the survey. Treatment duration is taken as the time from baseline visit to the last return visit recorded in the database. For the purposes of the present analysis, deaths and treatment before January 15, 1998, were considered. To ensure that all data before that date had been reported, the database was not analyzed until August 1998.

In the present analysis, 1903 patients were treated for 2334 yr. The mean treatment time was, thus 1.2 yr, with a maximum treatment of 4 yr. Sixty-three percent of the treatment time was beyond 6 months, and 35% was beyond 12 months. The patient sample represented 14 countries, and the mean age was 44 yr. Twenty-nine percent of the treatment time represented Swedish patients. The mean dose of GH was 1.5 IU (range, 0.2–6.4). To compare the mortality in KIMS to that from other sources, patients with a history of acromegaly or Cushing’s disease were excluded. The number of deaths in the remaining population was 11 (six males). The causes of death in these patients were cerebral aneurysm (2), myocardial infarction (2), cerebral bleeding (1), suicide (1), ischemic colitis (1), relapse of craniopharyngioma (1), adrenal insufficiency (1), and in two patients the cause of death was unknown. To investigate the quality of the reporting of deaths into the KIMS database, the Swedish subsample was controlled against the central Swedish population registry, and the two sources were found to be in complete concordance.

Expected mortality was obtained from sex and 10-yr age-specific death rates for each country (6). For countries representing 76% of the treatment time in KIMS the mortality data for comparison were from 1995. For countries representing half of the remaining time, the data were from 1994, and for half it was older. The data were, thus, fairly close to the midcalendar yr of KIMS. The expected number of deaths, from all causes, was calculated to be 11.7, and the SMR was 0.94. Based on Poisson distribution the 95% confidence interval (CI) was calculated to be 0.47–1.68.

The KIMS cohort described was also compared with the Swedish population of hypopituitary patients treated with conventional replacement therapy but not GH, on which the study by Rosén, was based. The data of Rosén et al. was restricted to the period 1980–1990. The risk ratio of this population vs. KIMS was calculated to be 3.25 when stratifying the comparison for sex and age, in 10-yr groups, but disregarding the difference in yr and country. A 95% CI for the risk ratio was calculated to be 1.64–8.07; the CI was based on the optimal, exact test for comparing two Poisson distributions.

The KIMS cohort of treated GHD patients consists mainly of patients with a previous pituitary tumor or craniopharyngioma. The cohort also has other unfavorable health characteristics, including hypertension and high lipids (5); the mortality might, thus, be expected to be higher than in the general population. However, because potential patients with an ongoing neoplasm are excluded from GH treatment, mortality due to neoplasia is expected to be low in this cohort, especially during initial treatment. The previously described calculations were, therefore, repeated, excluding deaths due to malignant diseases from all comparisons. The SMR was then 1.49, and the CI was 0.74–2.66. Comparing the data of Rosén et al. (2) of untreated GHD to the KIMS cohort, the risk ratio was 2.26 and the CI, 1.08–6.01.

Based on more than 2000 patient yr, we can conclude that GH replacement therapy to adult GHD patients is not associated with any increase in mortality. The doses of GH used in the ICU trials were approximately 10 times higher than the doses used in KIMS for replacement therapy, and clearly the doses used in the former may have had a major bearing on the adverse outcome. The lower mortality among the patients included into the KIMS database in comparison with the outcome of hypopituitary patients previously reported by Rosén et al. (2) could have several explanations, including selection of patients and the limited number of patients yr. Hence, it is too early to state that GH replacement therapy normalizes mortality in hypopituitary patients.

References

  1. Takala J, Ruokonen E, Webster NR, et al. 1999 Increased mortality associated with growth hormone treatment in critical ill adults. N Engl J Med. 341:785–792.[Abstract/Free Full Text]
  2. Rosén T, Bengtsson B-Å. 1990 Premature mortality due to cardiovascular disease in hypopituitarism. Lancet. 336:285–288.
  3. Bülow B, Hagmar L, Mikoczy Z, Nordstrom CH, Erfurth EM. 1997 Increased cerebrovascular mortality in patients with hypopituitarism. Clin Endocrinol 46:75–81.
  4. Bates AS, Bullivant B, Sheppard MC, Stewart PM. 1996 The effect of hypopituitarism on life expectancy. J. Clin. Endocrinol Metab. 81:1169–1172.
  5. Caroroll P, Christ R, Bengtsson B-Å, et al. 1998 Growth hormone deficiency in adulthood and the effects of growth hormone replacement: a review. J Clin Endocrinol Metab. 83:382–395.[Abstract/Free Full Text]
  6. WHO. 1997 World Health Statistics Annual, 1995 and 1996. Geneva: WHO.