Can immediate postoperative random growth hormone levels predict long-term cure in patients with acromegaly?
Correspondence Address: Source of Support: None, Conflict of Interest: None DOI: 10.4103/0028-3886.177622
Source of Support: None, Conflict of Interest: None
Background: Growth hormone (GH) levels following oral glucose tolerance test (OGTT) at 12 weeks or later after surgery have been accepted as the most reliable parameter for defining remission and/or cure in patients with acromegaly. However, the role of random GH in predicting remission in the immediate postoperative period using modern criteria is not known. This study was undertaken to evaluate the role of random GH levels in first 5 postoperative days as an early predictive tool for long-term remission of patients with acromegaly following transsphenoidal pituitary surgery (TSS).
Keywords: Acromegaly; long-term remission; postoperative random growth hormone; transsphenoidal surgery
Transsphenoidal surgery (TSS) is the treatment of choice for somatotropinoma, with dopamine agonist, somatostatin receptor ligands, growth hormone (GH) receptor antagonists, and radiotherapy being used as add-on therapies. After surgery, biochemical assessment is more important and reliable in predicting remission than magnetic resonance imaging (MRI).,,
Following surgery, biochemical monitoring of patients with acromegaly involves insulin-like growth factor-1 (IGF-1), nadir GH levels after an oral glucose tolerance test (OGTT) and single or multiple GH measurements. After the revised guidelines published in 2010, 2013, and 2014, the criteria for cure are normal levels of serum IGF-1 (adjusted for age) and either a nadir GH after an OGTT <0 .4 µg/l or random GH <1 µg/l at 12 weeks or later after surgery.,, A recent expert consensus statement on medical treatment of acromegaly suggested that nadir GH after OGTT may not be useful in the immediate postoperative period due to inconsistent results (though the quality of evidence was low)., In contrast to the situation in Cushing's disease, where immediate postoperative cortisol is an established parameter for predicting future remission, the role of immediate postoperative GH estimation for prediction of future remission is not established in acromegaly. As most of the authorities wait until 2–4 months following surgery for neuroimaging, from a clinical point of view, prediction of likelihood of cure immediately after surgery would be a valuable parameter. This was studied previously, mostly using older and less sensitive assays.,,,, In this study, we revisit this important issue using electrochemiluminescence GH assay, which is considered to be more sensitive and specific.
This prospective study was conducted between 2009 and 2014 at a tertiary endocrine referral center in North India. Written informed consent was taken from all patients, and the study was approved by the Institute's Ethics Committee. The study included 75 consecutive patients with acromegaly who underwent TSS in our institution. The diagnosis of acromegaly was based on clinical signs and symptoms and nonsuppressed GH after OGTT (75 g anhydrous glucose with sampling at 0, 30, 60, 90, and 120 min for GH estimation, cutoff for nadir ≥0.4 ng/ml) with elevated IGF-1 (matched for age). Contrast-enhanced MRI was performed in all patients. Patients who received radiotherapy or octreotide pre- or post-operatively were excluded from the study. All the surgeries were done by a single neurosurgeon (KKM) who has performed >100 TSS per year for the last 10 years. The preoperative pituitary hormone deficiencies were treated appropriately.
Samples for GH were taken immediately (as soon as patient regained consciousness from anesthesia) and at 6 h postsurgery, as well as daily from postoperative day 1 to 5. GH was measured by electrochemiluminescence (ECLIA, Cobas e602, Roche-Hitachi, MN, USA), with the lowest detection limit of 0.03 ng/ml and both inter- and intra-assay coefficient of variation being <2 .5%. The method has been standardized against international reference preparation control code 98/574. IGF-1 was measured by an automated growth hormone assay (DiaSorin, Liaison, Germany), with a lower detection limit of 25 ng/ml. Patients were followed for at least 1 year, during which multiple OGTT and IGF-1 were done. Cure was defined as normal IGF-1, a random GH of <1 ng/ml and nadir GH <0 .4 ng/ml during an OGTT at least at 3 months or later. Patients who did not meet these criteria were considered to have an active disease. Sera were stored at −80°C and all samples were run by a single assay to avoid variability. None of the patients were given high-dose steroid perioperatively. Only those patients who had documented glucocorticoid deficiency were given hydrocortisone infusion at 4 mg/h. Tumors were classified into giant adenoma (maximum tumor dimension of ≥4 cm) macroadenoma (>10 mm) or microadenoma (<10 mm) on the basis of MRI. Adenomas were also classified by extension according to Knosp's grading. Tumor volume was calculated by De Chiro and Nelson formula.
Baseline parameters were analyzed using chi-square test for binomial data and Mann–Whitney U-test or unpaired t-test for continuous variables, depending on whether the data were normally distributed or not. To determine the diagnostic value of various postoperative parameters, receiver operator characteristic (ROC) curves were constructed for the respective parameters using postoperative nadir GH <0 .4 ng/ml as the gold standard. Area under the curve (AUC) using the trapezoidal method was calculated and P < 0.05 was considered statistically significant. Statistical analysis was performed using SPSS 22 (IBM, Armonkny, NY, USA) and Minitab 16 (Minitab Inc., Pennsylvania, USA).
Seventy-five acromegaly patients (43 males) with multiple postoperative visits at 3–6 monthly intervals for at least 1 year after surgery with no adjuvant therapy were included for the final analysis. The mean age of the patients was 37 ± 2.1 years. The median duration of symptoms before surgery was 4 years (range: 2–8 years). On MRI, 68 patients had a macroadenoma (22 with supra and/or parasellar extension; 9 patients had a giant adenoma [≥4 cm]) while 7 had a microadenoma. Twenty-six patients had an invasive adenoma on combined radiological (n = 24) and histopathological criteria (n = 16). Eleven cured and 13 uncured Knosp grades 3 or 4 adenomas, as well as 2 cured microadenomas with dural invasion, were classified as invasive [Figure 1]. Visual field defects were present in 38 (50%) patients. At the time of analysis, patients were divided into two groups, namely the cured and uncured groups. During this period, patients who were uncured did not receive any other therapies. Demographic and biochemical characteristics are summarized in [Table 1]. During follow-up, 42 (56%) patients were cured whereas 33 (44%) were uncured. After completion of the study, uncured patients were treated with radiotherapy (n = 1), long-acting release octreotide (n = 3), and gamma-knife surgery (n = 4) and the remaining patients received cabergoline therapy due to cost constraints.
Preoperatively, there was no significant difference in GH levels between the groups. However, preoperative IGF-1 levels were significantly lower in patients who were cured. Before surgery, the median IGF-1 increase was 1.19 times (interquartile range [IQR] 0.68–2.61) of the upper limit of normal (ULN) in patients who were cured and 1.22 (IQR 0.75–2.72) in the patients who were not cured (P = 0.1). Three months following surgery, the median age-adjusted IGF-1 decreased to 0.58 of the ULN (IQR 0.45–0.79) in patients who were cured whereas in the uncured group, there was an increase to a median of 1.1 × ULN (IQR – 1.12 –1.45 × ULN, P < 0.001). On evaluation at 3 m, 12 patients (16%) had discordant results on GH and IGF-1 tests. Of these, 11 (14.7%) had nonsuppressible GH with low IGF-1, while 1 (1.3%) had suppressible GH with elevated IGF-1. In all of these patients, there was no change in the biochemical status. Patients with discordant results were closely followed up without any intervention. Even though IGF-1 values predicted the response, ROC analysis of increase in the age-adjusted IGF-1 did not predict cure.
The mean (±standard deviation) tumor volume was 4685 ± 6278.39 with no significant difference between the groups, the median volume in the cured group being 2576 mm 3 (range: 1192.3–6655.3); and, in the uncured group being 2520 mm 3 (1183–4657.5, P = 0.56). The tumor volume, Knosp's grading and invasion of bone, dura or mucosa were not predictive of long-term cure. The duration of disease before presentation was also not predictive of long-term cure (β = 0.96, P = 0.61). The median duration of disease in the cured patients was 4 years (2–10 years) whereas it was 3 years (2–5 years) for the uncured patients.
We evaluated the utility of postoperative GH levels at different time intervals for predicting long-term cure by analyzing ROC curves [Figure 2]. The preoperative GH levels showed a poor predictive value while the preoperative IGF-1 levels had a fair predictive value for long-term cure. GH levels in the immediate postoperative period, 6 h after surgery, and on postoperative day 1 to day 5 had a good predictive value for long-term cure. The highest predictive value was for GH levels 6 h postsurgery (0.892); and the lowest predictive value was on day 4 (0.8). Similarly, the postoperative IGF-1 levels at 3 months had an excellent predictive value (0.994) as well as percentage decrease in GH (0.908). In addition, >97% fall in GH value on day 1 as compared to the nadir preoperative GH value on OGTT was highly predictive of long-term remission. A similar result was found with basal GH levels on multivariate analysis (data not shown). ROC curves analyses for different parameters are enlisted in [Table 2].
We evaluated the sensitivity and specificity of GH levels before surgery, during the immediate postsurgery period, 6 h after surgery and on the postoperative day 1 to day 5, in predicting long-term cure in patients with acromegaly who underwent TSS. An immediate postsurgery GH level of 3.5 ng/ml had a very good sensitivity (>90%) and poor specificity (50%) whereas the sensitivity and specificity for a GH level of 1.5 ng/ml, 6 h postsurgery were 81.2% and 83.3%, respectively [Table 3]. A postoperative day 1 GH level of 1.03 ng/ml had a sensitivity of 90.5% and a specificity of 71.4% for predicting long-term cure whereas a postoperative day 2 GH level of 2.15 ng/ml had a sensitivity of 80% and a specificity of 75%. Positive predictive value for postoperative 6 h GH was 83% (cutoff is 1.5 ng/ml). A postoperative IGF-1 level of 155 ng/ml had a sensitivity of 100% for predicting long-term cure but had poor specificity (77.4%). Similarly, sensitivity and specificity of percentage fall in GH and IGF-1 levels for predicting long-term cure were also calculated.
Although the AUC for GH level (1.5 ng/ml) at 6 h and at day 1 (1.03 ng/ml) were comparable, the 1.5 ng/ml value at postoperative 6 h time point is closer to the left-hand side of the plot. Hence, this has a better sensitivity and specificity together rather than the day 1 values, which have only better sensitivity.
In one-third of patients (23), we could not correctly predict cured or uncured status with 6 h GH cutoff. Eighteen patients (24%) who initially had 6 h postoperative GH levels >1.5 ng/ml had normalized GH levels at 3 m. Five patients had 6 h GH <1 .5 ng/ml but on follow-up were found to have an active disease.
The aim of this study was to evaluate the sensitivity, specificity, and predictive value of the immediate postoperative GH levels in predicting long-term cure of patients with acromegaly who underwent TSS. The main finding of our study was that GH levels measured 6 h after surgery had the highest predictive value (AUC 0.892) for long-term cure. In addition, duration of the disease, the tumor volume, or Knosp's grading had no effect on the prediction of cure. Postoperative IGF-1 measured 3 months after surgery had a better predictive value (AUC = 0.99) than GH levels measured 6 h postsurgery (with threshold of 1.5 ng/ml, AUC = 0.89), but understandably this was not done in the immediate postoperative period.
In our study, although the immediate postoperative GH had a marginally higher sensitivity than the 6 h GH values, it had a poorer specificity as compared to the 6 h GH in predicting long-term cure. Postoperative IGF-1 level at 3 m of 0.8x of ULN had a sensitivity of 100% but specificity of only 77.8%. The percentage fall in IGF-1 and GH levels after surgery also had a good predictive value. A 6 h postoperative GH value below 1.5 ng/ml was a very sensitive and specific test to predict long-term cure. We had only 5 patients who had a 6 h GH value <1 .5 ng/ml but were not cured on subsequent follow-up. There were no clinical characteristics which could have identified these 5 patients. On the contrary, a GH value above this cutoff does not necessarily mean that patient is unlikely to be cured. Sixteen patients (38%) who were cured had a GH value >1.5 ng/ml 6 h after surgery.
Earlier studies had used radioimmunoassay (RIA) to measure GH levels. Yamada et al., and Freda et al., used the rapid RIA kit to estimate GH immediately after tumor resection, which has highlighted the ability of immediate postsurgery GH levels as well as extent of decrease in GH levels to predict the likelihood of cure.,,, Valdemarsson et al., estimated GH using immunoradiometric assay on postsurgery day 1 at 30 min intervals and random GH samples from postsurgery day 1 to day 5, to evaluate the utility of immediate postsurgery GH in predicting cure. Similarly, a study conducted by Kim et al., evaluated the utility of random postoperative GH levels in the first 3 days using a similar assay. They suggested that nadir GH after OGTT 1 week after surgery is a very sensitive modality in predicting cure; however, not only is it logistically difficult to call a patient within a couple of days of discharge, it is also cumbersome and more costly. Over the years, these older GH assays have been replaced by ECLIA and we used this technique, for the first time to assess the immediate postoperative GH levels. Recently, a study by Sarkar et al., used the same method; however, they estimated fasting plasma GH later, starting on postoperative day 1 and a nadir GH after OGTT on day 7. In their study, the suggested GH cutoffs were slightly higher (day 1 = 6.99 ng/ml and day 7 postglucose suppression = 5.54 ng/ml). As most of our patients are discharged by day 3–5, we wanted to develop a system for prediction of cure before discharge and this will further reduce the cost and duration of hospital stay.
As the GH assays improved over time, so did their lowest detection limits; therefore, the cutoffs for GH levels to predict cure have also been lowered. A recent expert consensus statement on the medical treatment of acromegaly suggested a random GH <1 ng/ml as a treatment goal after surgery, with a need to determine assay and method specific cutoffs. OGTT as a tool for follow-up gives inconsistent results and the study by Freda et al., identified a group of patients with normal age-adjusted IGF-1 and post-OGTT nadir GH of <1 ng/ml, where post-OGTT GH was higher than the level found in healthy controls (0.19 ng/ml). These patients had an increased chance of recurrence., The recent consensus statement also did not favor OGTT as a tool for follow-up due to poor reproducibility.,,,,
The use of different criteria to define biochemical remission; and, the variable duration of follow-up, accounts for differences in the overall surgical outcome in patients with acromegaly following TSS. The most recent surgical series report a cure rate of 30–60%, with the more stringent criteria for biochemical cure.,, In our study, the cure rate was 56% which is comparable to the recent studies.,, Although, the use of the current criteria reduces the immediate postoperative cure rate, the long-term recurrence rate is also markedly reduced. Therefore, radiological and histopathological criteria may not always be useful to predict cure. This is in contrast to the study reported by Laws and Trouillas et al., This could be explained by the completeness of surgery, absence of mitotic figures, and lack of correlation between the clinical, radiological, and biological behavior in general. However, none of our patients with an invasive adenoma had Trouillas grade 2B or 3. On the contrary, 5 patients (2 invasive) with the 6 h GH <1 .5 ng/ml were uncured on subsequent follow-up. Cure in this subset of patients could be an exception rather than a rule.
Determination of remission or cure in acromegaly is generally made several weeks or months after surgery on the basis of OGTT and IGF-1. It is based on the premise that IGF-1 has a long half-life and there is a progressive fall in the GH level following surgery due to ischemia and fibrosis of the tumor tissue. Early prediction of remission will boost the patient morale and avoid unnecessary radiological and endocrinological investigations. In addition, a subset of patients who are less likely to achieve cure may undergo repeat surgery during the same hospital stay, before significant cicatricial changes develop. The short half-life of GH potentially allows for prediction of an early endocrinological outcome. In patients with a somatotropinoma, GH hypersecretion from the adenomatous tissue leads to functional suppression of the surrounding, normal somatotroph cells. Therefore, the decline in postoperative GH level reflects the extent of surgical removal. In the present study, we evaluated the utility of random GH levels in the immediate postoperative period in predicting the likelihood of long-term cure in patients with acromegaly who underwent TSS. Early estimate of surgical failure will be useful to predict which patient may require early pharmacotherapy and/or radiotherapy/radiosurgery, and also help in deciding the intensity of postoperative follow-up.
The immediate postoperative GH levels can be used to predict long-term cure; and, the trends of serial GH values are useful for follow-up of patients after surgery. Our study showed the utility of immediate postsurgery GH levels as a useful tool for the prediction of long-term cure and in planning further treatment after TSS; however, a word of caution is essential, as in one-third of the patients, the immediate GH results may not predict the final outcome, casting doubts on the usefulness of this variable in clinical practice. The limitations of our study include a small sample size, which could have led to failure to find significant association between surgical outcome and some clinical characteristics due to type 2 error; a relatively shorter duration of follow-up and lack of comparison with OGTT before discharge (although we feel that performing OGTT in the immediate postoperative period is cumbersome and delays the hospital stay). In contrast to the situation in Cushing's disease, where assessing postoperative cure is of paramount importance in deciding an early re-intervention, the same may be less relevant in patients with acromegaly as the biochemical status is likely to change in a large proportion of patients in the first postoperative year.,
The immediate perioperative GH assessment is a predictor of remission in two-thirds of patients with acromegaly. If patients do not show an appropriate decrease in GH levels after surgery, the physician and surgeon could wait for a year before the final call for an alternate treatment strategy is made. Conversely, an early decision may also be taken in the immediate postoperative phase for the adjuvant plan.
We would like to acknowledge Dr. Ashu Rastogi (Assistant Professor, Department of Endocrinology, PGIMER, Chandigarh) for manuscript editing and constructive comments.
Financial support and sponsorship
Conflicts of interest
There are no conflicts of interest.
[Figure 1], [Figure 2]
[Table 1], [Table 2], [Table 3]