https://orcid.org/0000-0001-7810-8247
Abstract
A paradoxical growth hormone (GH) response to oral glucose load (oral glucose tolerance test [OGTT]) in acromegaly is associated with a milder phenotype.
To study whether the GH response to the OGTT predicts the risk of recurrence after initial surgical cure.
Retrospective, observational study at 2 tertiary care centers. We investigated 254 patients with acromegaly who were cured by surgery. All patients underwent an OGTT at diagnosis before pituitary surgery. A peak to basal GH ratio ≥120% within 90 minutes was used to distinguish paradoxical (GH-Par) from nonparadoxical responses (GH-NPar) in patients with acromegaly. Cox analysis was used to investigate whether the paradoxical GH response to OGTT was associated with the risk of disease recurrence
A paradoxical GH response to OGTT occurred in 87 patients (34.3%, termed GH-Par group). Recurrence of acromegaly occurred in 3 patients in the GH-Par group (3.4%) and in 20 patients in the GH-NPar group (12.0%). In the multivariate analysis, the paradoxical GH response to OGTT was significantly associated with a low risk of recurrence (HR 0.18, 95% CI 0.05-0.63; P = .007). Basal GH level at diagnosis was the only other variable associated with the risk of disease recurrence (HR 1.58, 95% CI 1.01-2.47; P = .04).
Our study demonstrates that a paradoxical GH response to OGTT in the preoperative setting predicts a lower risk of disease recurrence after initial surgical cure. The pattern of GH responsiveness to OGTT is, therefore, useful to predict the long-term outcome of patients with acromegaly.
Pituitary surgery by a skilled neurosurgeon represents the first-line treatment in the majority of patients with acromegaly (1, 2). In high-volume centers, the initial remission rate is approaching 85% for microadenomas and around 40% to 50% for macroadenomas (3-5). One caveat of pituitary surgery is recurrence of disease after an initially proven remission. Growth hormone (GH)–secreting pituitary adenomas are probably the least likely subtype of pituitary adenoma showing recurrence during long term follow-up (5, 6). Nevertheless, a recurrence rate of 2% to 8% is reported in large surgical series (3-5, 7), even though a recurrence rate as high as 31% occurred in smaller series (8, 9). Possible explanations for this heterogeneity are the length of follow-up (8), or the nonunivocal or rigorous definition of relapse/remission in different papers, variably based on insulin-like growth factor-1 (IGF-1) levels or GH values after oral glucose tolerance test (OGTT). It is plausible that adenoma recurrence arises from tiny tumor remnants left over at surgery that have limited hormone secretion and size to be detected in the early endocrine and/or radiological postoperative assessments (6). The search for useful predictors of disease recurrence is complicated because of the need to accrue a large cohort of patients with a long follow-up. Nevertheless, persistence of the paradoxical GH response to thyrotropin-releasing hormone administration (8), higher basal GH postoperative concentrations (10, 11), and GH suppression higher than 0.4 µg/L in the postsurgical OGTT assessment (12, 13) have been suggested to be predictive of disease recurrence.
The OGTT is useful for diagnosing active acromegaly (14). Besides its diagnostic purposes, the pattern of GH responses to OGTT in patients with acromegaly has gained widespread interest in recent years. Indeed, it has been shown that about one-third of patients show an increase in GH levels of more than 20% over baseline within the first 90 minutes from oral administration of glucose. The paradoxical increase in GH secretion after glucose load may have clinical and prognostic relevance (15-18). In particular, the presence of a paradoxical response to OGTT has been associated with a milder clinical phenotype, characterized by smaller and less invasive tumors, as well as a better response to first-generation somatostatin analogues (16, 17). It is, therefore, reasonable to postulate that the presence of a paradoxical GH response to OGTT might identify a subgroup of patients with biologically less aggressive pituitary tumors.
We tested the hypothesis that the pattern of GH responsiveness to OGTT might be associated with the risk of tumor recurrence after initial surgical remission.
Materials and Methods
Patients
The study was performed in 2 Italian referral centers for pituitary disease (Department of Neurosurgery, Vita-Salute University, Milan and Department of Endocrinology, University of Padua, Padua). We included consecutive patients with a diagnosis of active acromegaly who received first surgery in 1 of the 2 centers and who had postoperative clinical and biochemical remission of disease. Patients were also required to have the results of the OGTT at diagnosis and a minimal follow-up of 1 year after surgical remission.
Diagnosis of active acromegaly was confirmed by elevated GH and IGF-1 levels, failure of GH to suppress under 1 µg/L during OGTT, a magnetic resonance imaging demonstrating a pituitary adenoma, and histological confirmation of a GH-secreting pituitary adenoma. Surgical remission of acromegaly was defined as normalization of age- and sex-adjusted IGF-1 concentration together with a suppressed serum GH level during OGTT, according to the guidelines used in clinical practice at the moment of the evaluation (1). Patients receiving medication(s) before surgery were required to fulfil the above-mentioned criteria at least 6 months after surgery to avoid a possible carryover effect of medications.
Recurrence of acromegaly during follow-up was confirmed by increased age- and sex-adjusted IGF-1 levels, as well as elevated basal GH and/or lack of suppression during OGTT, according to the guidelines and assay methodology in use at the time of evaluation (1).
Standard informed consent for the collection of data and surgical procedures was obtained from all the patients. The study protocol was approved by the ethics committee of both hospitals (Milan: approval number NCH-01-21; Padova: PP-PIT: Predictive and Prognostic factors in PITuitary Adenoma, AOP1782, CESC 4834/AO/20).
OGTT and Criteria of Response
All patients underwent an OGTT for diagnostic purpose at presentation and before commencing any GH-lowering medication. The GH sampling was performed after an overnight fast and following a standard protocol. Briefly, blood specimens were taken before and then 30, 60, 90, and 120 minutes after oral administration of 75 g of glucose. Serum basal GH was defined as the fasting serum GH immediately prior to glucose administration. The GH responsiveness to the OGTT was evaluated as the peak to basal GH ratio (OGTT ratio). In keeping with our previous publications (16, 19, 20), an OGTT ratio ≥120% at latest 90 minutes after glucose load was used to distinguish paradoxical (GH-Par) from nonparadoxical patients (GH-NPar).
Hormone Measurement
Whenever feasible, GH and IGF-1 levels were measured at each center by using commercially available laboratory kits (20, 21). In Milan and Padua, GH levels were measured with Immulite 2000 (Siemens Healthcare, Diagnostics Products Ltd, Llanberis, Gwynedd, UK; RRID:AB_2811291). The GH assay has an analytical measurement range of 0.05 to 40 μg/L with an intraassay coefficient of variation less than 4% at low (3.28 μg/L), mid (8.25 μg/L), and high (13.00 μg/L) GH levels.
Serum IGF-1 levels were measured in Milan with an automated immunometric assay (Immulite 2000, Siemens Healthcare, Diagnostics Products Ltd, Llanberis, Gwynedd, UK; RRID:AB_2756880) and in Padua with another automated immunometric assay (Liaison IGF-I, Diasorin, Saluggia, Italy; RRID:AB_2928957).
However, most patients either at diagnosis or during follow-up had hormonal measurements performed with different assay kits. For this reason, IGF-1 values are presented as a multiple of the upper limit of normal (mULN) for age and sex.
Statistical Analysis
The primary endpoint of the study was to investigate whether the paradoxical GH response to OGTT was associated with the risk of disease recurrence after surgical remission. A sensitivity analysis using a more restrictive criterion (ie, an OGTT ratio ≥150% to classify GH-Par and GH-NPar patients) was also performed.
Continuous data are expressed as mean ± SE of the mean or median values and interquartile range (IQR), as appropriate. The Shapiro–Wilks test was performed to test for normality. The Student's t-test was used to identify mean differences between groups in the case of Gaussian-like variables, otherwise the Mann–Whitney test was used. The Pearson's chi-squared test was applied to categorical data. The cumulative risk of tumor recurrence was calculated according to the Kaplan–Meier method and the log rank test. Recurrence-free survival was measured starting from the time of the surgical procedure. Data on tumor recurrence were censored at the time of the last hormonal follow-up. The association of the GH paradoxical response to the OGTT as well as of other clinical characteristics with the risk of tumor recurrence was explored with the use of a Cox proportional hazard model and expressed as hazard ratio (HR) and 95% CI. The final multivariate model included all the variables with a P value less than .05 in the univariate analysis. A probability value less than .05 was considered to be statistically significant, and all reported analyses are 2-tailed. All calculations were performed using IBM SPSS Statistics, version 23.0 (IBM Corp., Armonk, NY, USA).
Results
Patients” Characteristics
The study group included 254 patients with active acromegaly who were followed for at least 1 year after successful removal of a GH-secreting pituitary adenoma. Table 1 summarizes the characteristics of the patients included into the study. There was a small preponderance of females and the mean age at surgery was 47.6 ± 0.8 years. Seventy patients (27.6%) had a microadenoma and only 4.7% and 5.5% had invasion of the cavernous sinus and sphenoid sinus, respectively.
Baseline clinical characteristics of 254 patients with acromegaly who participated into the study
| Variables | All (n = 254) | GH-par (n = 87) | GH-Npar (n = 167) | P value |
|---|---|---|---|---|
| Age at surgery, years (mean ± SE) | 47.6 ± 0.8 | 49.1 ± 1.3 | 46.8 ± 0.9 | .16 |
| Male sex, n (%) | 117 (46.1) | 46 (52.9) | 71 (42.5) | .15 |
| Basal GH level, µg/L (median, IQR) | 9.0 (4.6-16.2) | 9.3 (4.4-15.8) | 9.0 (4.6-16.7) | .89 |
| Basal IGF-1, mULN (mean ± SEM) | 2.92 ± 0.10 | 3.13 ± 0.14 | 2.76 ± 0.14 | .07 |
| Microadenoma, n (%) | 70 (27.6) | 28 (32.2) | 42 (25.1) | .30 |
| Maximum tumor diameter, mm (mean ± SEM)a | 14.8 ± 0.4 | 13.4 ± 0.6 | 15.5 ± 0.6 | .02 |
| Cavernous sinus invasion, n (%) | 12 (4.7) | 3 (3.4) | 9 (5.4) | .77 |
| Sphenoid sinus invasion, n (%) | 14 (5.5) | 4 (4.6) | 10 (6.0) | .55 |
| Variables | All (n = 254) | GH-par (n = 87) | GH-Npar (n = 167) | P value |
|---|---|---|---|---|
| Age at surgery, years (mean ± SE) | 47.6 ± 0.8 | 49.1 ± 1.3 | 46.8 ± 0.9 | .16 |
| Male sex, n (%) | 117 (46.1) | 46 (52.9) | 71 (42.5) | .15 |
| Basal GH level, µg/L (median, IQR) | 9.0 (4.6-16.2) | 9.3 (4.4-15.8) | 9.0 (4.6-16.7) | .89 |
| Basal IGF-1, mULN (mean ± SEM) | 2.92 ± 0.10 | 3.13 ± 0.14 | 2.76 ± 0.14 | .07 |
| Microadenoma, n (%) | 70 (27.6) | 28 (32.2) | 42 (25.1) | .30 |
| Maximum tumor diameter, mm (mean ± SEM)a | 14.8 ± 0.4 | 13.4 ± 0.6 | 15.5 ± 0.6 | .02 |
| Cavernous sinus invasion, n (%) | 12 (4.7) | 3 (3.4) | 9 (5.4) | .77 |
| Sphenoid sinus invasion, n (%) | 14 (5.5) | 4 (4.6) | 10 (6.0) | .55 |
Patients were further subdivided in 2 groups according to the GH response to OGTT.
Abbreviations: GH, growth hormone; IGF-1, insulin-like growth factor 1; IQR, interquartile range; mULN, multiple of the upper limit of the normal; SE, standard error of the mean.
aInformation on maximum tumor diameter was not available for 5 patients (2 in the GH-Par group and 3 in the GH-NPar group)
Baseline clinical characteristics of 254 patients with acromegaly who participated into the study
| Variables | All (n = 254) | GH-par (n = 87) | GH-Npar (n = 167) | P value |
|---|---|---|---|---|
| Age at surgery, years (mean ± SE) | 47.6 ± 0.8 | 49.1 ± 1.3 | 46.8 ± 0.9 | .16 |
| Male sex, n (%) | 117 (46.1) | 46 (52.9) | 71 (42.5) | .15 |
| Basal GH level, µg/L (median, IQR) | 9.0 (4.6-16.2) | 9.3 (4.4-15.8) | 9.0 (4.6-16.7) | .89 |
| Basal IGF-1, mULN (mean ± SEM) | 2.92 ± 0.10 | 3.13 ± 0.14 | 2.76 ± 0.14 | .07 |
| Microadenoma, n (%) | 70 (27.6) | 28 (32.2) | 42 (25.1) | .30 |
| Maximum tumor diameter, mm (mean ± SEM)a | 14.8 ± 0.4 | 13.4 ± 0.6 | 15.5 ± 0.6 | .02 |
| Cavernous sinus invasion, n (%) | 12 (4.7) | 3 (3.4) | 9 (5.4) | .77 |
| Sphenoid sinus invasion, n (%) | 14 (5.5) | 4 (4.6) | 10 (6.0) | .55 |
| Variables | All (n = 254) | GH-par (n = 87) | GH-Npar (n = 167) | P value |
|---|---|---|---|---|
| Age at surgery, years (mean ± SE) | 47.6 ± 0.8 | 49.1 ± 1.3 | 46.8 ± 0.9 | .16 |
| Male sex, n (%) | 117 (46.1) | 46 (52.9) | 71 (42.5) | .15 |
| Basal GH level, µg/L (median, IQR) | 9.0 (4.6-16.2) | 9.3 (4.4-15.8) | 9.0 (4.6-16.7) | .89 |
| Basal IGF-1, mULN (mean ± SEM) | 2.92 ± 0.10 | 3.13 ± 0.14 | 2.76 ± 0.14 | .07 |
| Microadenoma, n (%) | 70 (27.6) | 28 (32.2) | 42 (25.1) | .30 |
| Maximum tumor diameter, mm (mean ± SEM)a | 14.8 ± 0.4 | 13.4 ± 0.6 | 15.5 ± 0.6 | .02 |
| Cavernous sinus invasion, n (%) | 12 (4.7) | 3 (3.4) | 9 (5.4) | .77 |
| Sphenoid sinus invasion, n (%) | 14 (5.5) | 4 (4.6) | 10 (6.0) | .55 |
Patients were further subdivided in 2 groups according to the GH response to OGTT.
Abbreviations: GH, growth hormone; IGF-1, insulin-like growth factor 1; IQR, interquartile range; mULN, multiple of the upper limit of the normal; SE, standard error of the mean.
aInformation on maximum tumor diameter was not available for 5 patients (2 in the GH-Par group and 3 in the GH-NPar group)
A paradoxical increase of GH after OGTT occurred in 87 patients (34.3%), while the remaining 167 patients (65.7%) did not show such a response. In the GH-Par group the median peak GH after OGTT was 163% higher than the basal GH level (IQR 135-200%, range 120-1228%). In the GH-NPar group, 120 patients (71.9%) showed a substantial decrease of GH levels during OGTT (ie, OGTT ratio, 80% or less of its basal level), while the remaining 47 patients (28.1%) had minor GH fluctuations (between 81% and 119% of the basal level). Table 1 summarizes the principal characteristics of the 2 groups of patients. Age at surgery, gender, basal GH level, IGF-1 mULN, presence of a microadenoma, and invasion of the cavernous sinus or the sphenoid sinus were similar between GH-Par and GH-NPar patients. In contrast, the maximum tumor diameter in the GH-Par group was significantly smaller than in the GH-NPar (Table 1).
When using the more restrictive criterion of an OGTT ratio of at least 150% to define a paradoxical GH response to OGTT, we classified 52 patients (20.5%) as GH-Par and the remaining 202 patients (79.5%) as GH-NPar.
Recurrence of Acromegaly
The median biochemical follow-up in the whole group of patients was 77 months (IQR 29-120 months, range 12-306 months).
During the study period, recurrence of acromegaly occurred in 23 patients (9.1%). The 5- and 10-year recurrence-free survivals of the whole study population were 94.3% (95% CI 90.9-97.7%) and 87.0% (95% CI 81.0-93.0%), respectively (Fig. 1). Most of the recurrences of disease occurred within 10 years of surgery.
Kaplan–Meier analysis showing the recurrence-free survival in 254 patients with acromegaly after surgical remission of disease. The 5- and 10-year recurrence-free survivals were 94.3% (95% CI 90.9-97.7%) and 87.0% (95% CI 81.0-93.0%), respectively.
Recurrence of disease occurred in only 3 patients of the GH-Par group (3.4%), while it occurred in 20 patients in the GH-Npar (12.0%). Kaplan–Meier analysis confirmed that the risk of recurrence in the 2 groups of patients was significantly different (P = .003 by the log-rank test; Fig. 2). The 5- and 10-year recurrence-free survival in the GH-Par group were 100% and 96.4% (95% CI 91.4-99.4%), respectively, compared with 90.9% (95% CI 85.7-96.1%) and 80.5% (95% CI 71.1-89.9%), respectively, in the GH-NPar group (Fig. 2).
Kaplan–Meier analysis showing the recurrence-free survival in 87 patients who had a paradoxical GH response to OGTT (continuous line) and in 167 patients who did not have such a response (dashed line). The difference between the 2 groups was significant (P = .003 by the log-rank test).
Univariate Cox analysis demonstrated that age at surgery, maximum tumor diameter, basal GH level at diagnosis (log transformed), and the pattern of GH response to OGTT were significantly associated with the risk of disease recurrence (Table 2), while gender, IGF-1 levels at diagnosis, cavernous sinus invasion, and sphenoid sinus invasion were not. We next performed a multivariate Cox analysis including all the variables that were significant in the univariate analysis. The final model included 249 patients with a complete dataset. The GH-Par group had a significant lower risk of developing recurrence of disease during follow-up than the GH-Npar group (HR 0.18, 95% CI 0.05-0.63; P = .007; Table 2). Basal GH level at diagnosis was the only other variable associated with the risk of disease recurrence (HR 1.58, 95% CI 1.01-2.47; P = .04), whereas age at surgery and maximum tumor diameter were no longer significant (Table 2).
Unadjusted and adjusted hazard ratio for the risk of recurrence of disease in 254 patients after surgical remission of acromegaly
| Covariate | Unadjusted HR (95% CI) | P value | Adjusted HR (95% CI)a | P value |
|---|---|---|---|---|
| Age at surgery (years) | 0.95 (0.92-0.99) | .01 | 0.97 (0.93-1.01) | .11 |
| Gender (female vs male) | 1.76 (0.74-4.16) | .20 | — | — |
| Maximum tumor diameter (mm)b | 1.08 (1.03-1.13) | .001 | 1.05 (0.99-1.11) | .08 |
| GH at diagnosis (log transformed) | 1.81 (1.23-2.66) | .002 | 1.58 (1.01-2.47) | .04 |
| IGF-1 at diagnosis (mULN) | 0.77 (0.42-1.39) | .38 | — | — |
| Cavernous sinus invasion (yes vs no) | 0.80 (0.11-5.98) | .80 | — | — |
| Sphenoid sinus invasion (yes vs no) | 1.41 (0.32-6.16) | .64 | — | — |
| GH response to OGTT (GH-Par vs GH-NPar) | 0.19 (0.06-0.64) | .008 | 0.18 (0.05-0.63) | .007 |
| Covariate | Unadjusted HR (95% CI) | P value | Adjusted HR (95% CI)a | P value |
|---|---|---|---|---|
| Age at surgery (years) | 0.95 (0.92-0.99) | .01 | 0.97 (0.93-1.01) | .11 |
| Gender (female vs male) | 1.76 (0.74-4.16) | .20 | — | — |
| Maximum tumor diameter (mm)b | 1.08 (1.03-1.13) | .001 | 1.05 (0.99-1.11) | .08 |
| GH at diagnosis (log transformed) | 1.81 (1.23-2.66) | .002 | 1.58 (1.01-2.47) | .04 |
| IGF-1 at diagnosis (mULN) | 0.77 (0.42-1.39) | .38 | — | — |
| Cavernous sinus invasion (yes vs no) | 0.80 (0.11-5.98) | .80 | — | — |
| Sphenoid sinus invasion (yes vs no) | 1.41 (0.32-6.16) | .64 | — | — |
| GH response to OGTT (GH-Par vs GH-NPar) | 0.19 (0.06-0.64) | .008 | 0.18 (0.05-0.63) | .007 |
Abbreviations: GH, growth hormone; GH-Par, GH paradoxical; GH-NPar, GH nonparadoxical; HR, hazard ratio; IGF-1, insulin-like growth factor 1; mULN, multiple of upper limit of normal; OGTT, oral glucose tolerance test.
aHazard ratios were adjusted for the covariates that were significant in the unadjusted analysis. The final model included 249 patients due to missing information in 1 variable.
bInformation on maximum tumor diameter was not available for 5 patients (2 in the GH-Par group and 3 in the GH-NPar group).
Unadjusted and adjusted hazard ratio for the risk of recurrence of disease in 254 patients after surgical remission of acromegaly
| Covariate | Unadjusted HR (95% CI) | P value | Adjusted HR (95% CI)a | P value |
|---|---|---|---|---|
| Age at surgery (years) | 0.95 (0.92-0.99) | .01 | 0.97 (0.93-1.01) | .11 |
| Gender (female vs male) | 1.76 (0.74-4.16) | .20 | — | — |
| Maximum tumor diameter (mm)b | 1.08 (1.03-1.13) | .001 | 1.05 (0.99-1.11) | .08 |
| GH at diagnosis (log transformed) | 1.81 (1.23-2.66) | .002 | 1.58 (1.01-2.47) | .04 |
| IGF-1 at diagnosis (mULN) | 0.77 (0.42-1.39) | .38 | — | — |
| Cavernous sinus invasion (yes vs no) | 0.80 (0.11-5.98) | .80 | — | — |
| Sphenoid sinus invasion (yes vs no) | 1.41 (0.32-6.16) | .64 | — | — |
| GH response to OGTT (GH-Par vs GH-NPar) | 0.19 (0.06-0.64) | .008 | 0.18 (0.05-0.63) | .007 |
| Covariate | Unadjusted HR (95% CI) | P value | Adjusted HR (95% CI)a | P value |
|---|---|---|---|---|
| Age at surgery (years) | 0.95 (0.92-0.99) | .01 | 0.97 (0.93-1.01) | .11 |
| Gender (female vs male) | 1.76 (0.74-4.16) | .20 | — | — |
| Maximum tumor diameter (mm)b | 1.08 (1.03-1.13) | .001 | 1.05 (0.99-1.11) | .08 |
| GH at diagnosis (log transformed) | 1.81 (1.23-2.66) | .002 | 1.58 (1.01-2.47) | .04 |
| IGF-1 at diagnosis (mULN) | 0.77 (0.42-1.39) | .38 | — | — |
| Cavernous sinus invasion (yes vs no) | 0.80 (0.11-5.98) | .80 | — | — |
| Sphenoid sinus invasion (yes vs no) | 1.41 (0.32-6.16) | .64 | — | — |
| GH response to OGTT (GH-Par vs GH-NPar) | 0.19 (0.06-0.64) | .008 | 0.18 (0.05-0.63) | .007 |
Abbreviations: GH, growth hormone; GH-Par, GH paradoxical; GH-NPar, GH nonparadoxical; HR, hazard ratio; IGF-1, insulin-like growth factor 1; mULN, multiple of upper limit of normal; OGTT, oral glucose tolerance test.
aHazard ratios were adjusted for the covariates that were significant in the unadjusted analysis. The final model included 249 patients due to missing information in 1 variable.
bInformation on maximum tumor diameter was not available for 5 patients (2 in the GH-Par group and 3 in the GH-NPar group).
In the sensitivity analysis, using the alternative criterion of an OGTT ratio of at least 150%, no patients of the GH-Par group had recurrence of disease during follow-up, while all 23 recurrences occurred in the 202 patients classified as GH-NPar (11.4%; P = .006).
Discussion
We found that patients with a paradoxical GH response to OGTT had a lower risk of disease recurrence after surgical remission than patients who did not have the paradoxical response. OGTT is, therefore, not only useful for diagnostic purposes but also as a predictive tool for the risk of tumor recurrence.
The percentage of GH-Par patients in our study is in keeping with that found in other unselected series of patients with acromegaly (15-18), but some characteristics, such as the high proportion of microadenomas and the low prevalence of cavernous sinus invasion are clearly due to the inclusion only of patients cured by surgery.
Despite an apparently lower risk of disease recurrence in GH-secreting pituitary adenoma than in other types of pituitary adenomas (5, 6), the overall recurrence rate in our cohort of patients was 9.1%, slightly higher than the 4.9% rate reported in 1 meta-analysis (6). Recurrences of disease occurred progressively in the first 10 years of follow-up and then were unusual. The same type of temporal distribution, slightly different from that of adrenocorticotropin- and prolactin-secreting pituitary adenomas, was observed in the meta-analysis (6). The quest for identifying simple and accurate preoperative and postoperative characteristics associated with the long-term prognosis of surgically treated patients with pituitary adenomas has been a research priority for several years. This task is arduous because of the small number of recurrences in most series, which limits the power of the statistical analysis. Not unexpectedly, clinical or hormonal characteristics associated with the risk of acromegaly recurrence have rarely and not consistently been reported so far (8, 10-13). Coopmans and coworkers (22) recently reported a Dutch multicenter study suggesting that larger tumor size and higher basal GH level at diagnosis were significant predictors of long-term recurrence of disease (22). However, only 9 patients had a recurrence, thus limiting the statistical power of the analysis (22). Given this background, the results of our study are noteworthy because the sample size of 254 patients allowed us to accrue an adequate number of recurrences (23 cases) for a meaningful analysis. Univariate analysis indicated that younger age, greater maximum diameter of the tumor, higher GH level at diagnosis, and the absence of a paradoxical GH response to OGTT were predictive of the risk of disease recurrence. However, multivariate analysis unequivocally showed that the GH response to OGTT remained the most important independent predictive factor, even though the baseline GH level at diagnosis was also a significant variable (Table 2). The strength of the association was further reinforced by the sensitivity analysis that used a more restrictive cut off to classify the paradoxical GH response to an OGTT. In this analysis, no patient who had an OGTT ratio ≥150% had recurrence of disease. Another single study investigated the possible association of the GH response to OGTT and the risk of recurrence (18) without finding any significant association. The main limitations of that study were the small sample size (67 patients available for the analysis of recurrence) and the relatively high rate of recurrence (11 of 67 patients, 16.4%), which is in contrast with the lower risk of disease recurrence reported in most studies (6). Moreover, the definition and timing of recurrence were lacking (18).
The mechanism(s) underpinning the better long-term outcome of GH-Par patients is unknown. We can only speculate that the lower risk of tumor recurrence is due to an intrinsically biological milder trait of the tumor in GH-Par patients. Indeed, GH-Par patients have been found to be older, to have smaller tumor size, and a lower incidence of cavernous sinus invasion than GH-NPar patients (16, 18, 23). Even more interestingly, GH-par patients also had a better responsiveness to first-generation somatostatin analogues administration than GH-NPar patients (16, 18, 24). From a molecular and histological standpoint, we have demonstrated that the paradoxical increase in GH during OGTT is primarily attributed to the ectopic activation of the GIP/GIPR axis at the somatotropinoma level (19, 25). Furthermore, Hage and colleagues (26) reported a higher incidence of the paradoxical GH response to an OGTT in tumors with elevated copy number alterations, which were also more frequently densely granulated. Findings on cytokeratin staining patterns, albeit based on a limited number of cases, supported this observation (16). Recently, Chasseloup and coworkers demonstrated that KDM1A haploinsufficiency was more frequent in GIPR-expressing GH-secreting adenomas, suggesting that decreased KDM1A function likely causes transcriptional derepression of the GIPR locus (27). All these data together support the notion that a different GH profile during OGTT might reflect diverse biological features of the tumor, with those exhibiting a paradoxical increase in GH after glucose challenge having a more favorable clinical phenotype.
The main limitations of our study lie in its observational and retrospective design. However, the large number of patients may partially overcome this limitation. Acromegaly is a rare condition and recurrence of disease is an infrequent event: performing a prospective, randomized study, involving the needed number of patients for analysis is, therefore, almost impossible. Currently, there is no consensus regarding the criteria for defining a paradoxical GH response pattern. Most studies considered a 20% to 30% increase as the arbitrary threshold to classify the GH response to OGTT (16-18, 20), but other authors considered a 50% increase for the definition of the paradoxical response (23, 28). Recognizing the arbitrary nature of the cut off value, we also performed a sensitivity analysis, setting a GH increase of at least 50% as the classification threshold. The results using this more restrictive criterion were even more impressive, as no patients in the GH-Par group had recurrence of disease, thus diminishing the possibility that the association we demonstrated was a chance finding secondary to misclassification of the GH response to OGTT. Another limitation of a retrospective study covering almost 30 years of data accrual is the heterogeneity of hormonal data. The criteria for surgical cure and recurrence of acromegaly have changed over the years because of the different methodologies to measure GH and IGF-1 levels. However, we have classified surgical results in keeping with the methodology in use at the time of evaluation. The recurrence rate we found is completely consistent with that reported in the literature (6). Moreover, assignment of recurrence during the follow-up period was done when the possible link to the GH response to OGTT was not yet hypothesized. Nevertheless, the possibility of misclassifying early surgical results and the consequent recurrence rate cannot be completely dismissed, but this is unlikely to affect the results of our study, because the supposed misclassification would have occurred at random in the GH-Par and GH-NPar groups. The requirement of at least 1 year of follow-up after surgery should minimize the risk, especially in patients who received somatostatin analogs before surgery, to misclassify early surgical results, and, hence, to count as true recurrence of disease cases who actually were not cured by surgery. Lastly, our study lacks data on radiological and histological parameters possibly associated with the biological behavior of the GH-secreting pituitary adenoma, such as T2-signal intensity on magnetic resonance imaging, granulation pattern, and proliferation index. The possibility that the association between the GH responsiveness to an OGTT and the risk of tumor recurrence might be mediated by 1 or more of these unmeasured variables cannot be excluded at present. However, classification of the GH response to the OGTT is an easy and cheap variable to obtain when trying to assess the long-term risk of recurrence of disease.
In conclusion, our study is the first to provide solid evidence that a paradoxical GH response to an OGTT in the preoperative setting is predictive of a lower risk of disease recurrence after initial surgical cure. The precise mechanism remains uncertain, but several observations support the hypothesis that patients in the GH-Par group have tumors with favorable underlying biological characteristics. The pattern of GH responsiveness to an OGTT is, therefore, useful to predict the long-term outcome of patients with acromegaly.
Disclosures
The authors declare that they have no conflict of interest.
Data Availability
Some or all data sets generated during and/or analyzed during the present study are not publicly available but are available from the corresponding author on reasonable request.
References
Abbreviations
- GH
growth hormone
- GH-NPar
nonparadoxical growth hormone response
- GH-Par
paradoxical growth hormone response
- IGF
insulin-like growth factor
- mULN
multiple of the upper limit of normal
- OGTT
oral glucose tolerance test
