Hypopituitarism after Tuberculous Meningitis in Childhood

Methods

The outpatient records of all patients treated at the Ruttonjee Hospital for tuberculous meningitis were reviewed. All patients who had tuberculous meningitis before the age of 21 years and had completed antituberculosis treatment at least 1 year before the study were invited to participate in this retrospective study. Tuberculous meningitis was defined as meningitis proved by lymphocytosis in the cerebrospinal fluid, with or without a positive culture for mycobacterium tuberculosis and which had responded to treatment with antituberculosis drugs. Of 246 patients who fulfilled the above inclusion criteria, 172 could not be contacted because of changes in address; all of these patients had discontinued follow-up at least 5 years previously because they had no residual neurologic dysfunction or epilepsy requiring medical attention. Of the 74 patients who could be contacted by phone or mail, 20 refused to participate in the study because they did not feel that they required any medical assessment; another 5 were so disabled that their relatives refused to bring them to the hospital for further investigation. A total of 49 patients consented to undergo de-tailed endocrine assessment at the Endocrine Division, Queen Mary Hospital. The participants included 27 men and 22 women, ages 23.4 6.0 years (mean SD), who had tuberculous meningitis at the age of 5.9 5.0 years, 17.5 5.2 years before the study.

Endocrine Assessment

In addition to clinical assessment for hypopituitarism by an endocrinologist, hypothalamicpituitary function was tested in all patients. The patients were studied at 0830 hours after an overnight fast. At least two basal samples were taken for serum and urinary osmolality and serum luteinizing hormone, follicle-stimulating hormone, prolactin, thyrotropin (TSH), estradiol or testosterone, and thyroxine (T4) measurements. The results were given as the mean of the two basal samples. On the same day, serum growth hormone and plasma cortisol responses to an insulin (0.15 u/kg body weight intravenously) tolerance test (nadir of plasma glucose 2.2 mmol/L), the serum gonadotropin responses to gonadotropin releasing hormone (GnRH, 100 g intravenously), and the serum TSH responses to thyrotropin releasing hormone (200 g intravenously) were assessed as previously described [6, 7]. In patients with epilepsy, for whom the insulin tolerance test was contraindicated, growth hormone and cortisol responses to glucagon (1 mg subcutaneously) stimulation were assessed [8]. For those with subnormal growth hormone or cortisol responses to glucagon, further assessment was done using the L-dopa-propranolol stimulation test [8] or short synacthen test [9], respectively. For patients with subnormal growth hormone or cortisol responses to the above stimulation tests, growth hormone or cortisol responses to growth hormone releasing hormone (100 g intravenously) or ovine corticotropin releasing hormone (100 g intravenously) were assessed within 6 months, as previously described [10]. Corticotropin (ACTH) responses to corticotropin releasing hormone were also measured [11]. In patients with symptoms or osmolality results suggestive of diabetes insipidus, a water deprivation test was done on another day.

Normal growth hormone and ACTH reserves were defined by a peak growth hormone level of 7.5 g/L or more [12] and a peak cortisol level of 550 nmol/L or more during any of the stimulation tests. The normal basal levels for prolactin, TSH, T4, and testosterone were 22.5 g/L, 7 mU/L or less, 62 to 154 nmol/L, and 10 to 40 nmol/L, respectively. Normal thyrotroph function was indicated by the presence of normal T4 and TSH levels. A normal hypothalamicpituitarygonadal axis was indicated by normal estradiol levels in a menstruating woman with normal prolactin levels or, in men, by the presence of normal testosterone and gonadotropin levels.

Analytic Methods

Serum growth hormone, ACTH, and prolactin were measured by two-site immunoradiometric assay (IRMA) using Allegro HS-ACTH IRMA kit (Nichols Diagnostics, San Juan Capistrano, California), Allegro H-GH IRMA kit (Nichols), and Abbott Prolactin Riabead Diagnostic kit (Abbott Laboratories, North Chicago, Illinois), respectively. Plasma cortisol and serum T4 were measured by fluorescence polarization immunoassay using kits from Abbott. Serum TSH was assayed using TSH enzyme immunoassay kits (Abbott). Serum gonadotropins were measured with Amerlex-M RIA Kit (Amersham, England). Serum testosterone and estradiol were measured by radioimmunoassay with World Health Organization (WHO) Matched Assay Reagents, using WHO standard batches K0789 and K0769 for testosterone and estradiol, respectively.

Magnetic Resonance Imaging Studies

All MRI scans studies were done and interpreted by one of the investigators at St. Teresa's Hospital. All examinations were conducted using Siemens model Magnetom SP 1-tesla unit (Erlangen, Germany). Images included pre-gadolinium T1-weighted coronal scans and post-gadolinium T1-weighted sagittal, axial, and coronal scans. The repetition time and echo time were 500 to 750 ms and 15 ms, respectively. The dose of gadolinium injected was 0.1 mmol/kg body weight. Except for the sagittal sequences, which were obtained with one acquisition at 4.5 to 4.8 mm intervals, all sequences were done with three to four acquisitions at 3- to 4-mm intervals.

Statistical Analysis

The Student unpaired t-test was used for the comparison of continuous variables. The Fisher exact test was used to compare all other variables. Those variables that were found to have a P value 0.1 were reassessed using multivariate stepwise discriminant analysis. Height standard deviation score (SDS) is calculated as follows [13]:

X-X

SDS =-

SD

Where X is the patient's height and X and SD are the population mean height and standard deviation, respectively, at that age.

Results

Endocrine Dysfunction

Of the 49 patients studied, only two had previously sought medical advice for possible endocrine problems. These two patients, who complained of primary amenorrhea and severe oligomenorrhea, respectively, had evidence of hypogonadotropic hypogonadism. Eight other patients were also found to have abnormal pituitary function (Table 1). A water deprivation test was considered to be indicated in eight patients; none had evidence of diabetes insipidus.

Gonadotropin responses to GnRH (Table 2) were normal in three of the five patients with gonadotropin deficiency, and were subnormal in the other two (patients 3 and 4). Because one patient (patient 4) had associated tuberculous endometritis, she had no withdrawal bleeding after estrogen and progestogen replacement. Growth hormone deficiency was the most common abnormality detected and was found in seven patients. The adult height and growth hormone responses to conventional stimulation tests and growth hormone releasing hormone in these patients are summarized in Table 3. In one patient (patient 2) who was studied 3 years after tuberculous meningitis, a large discrepancy was found between the growth hormone responses to hypoglycemia and growth hormone releasing hormone. In the other four patients who had tuberculous meningitis 18 to 36 years previously, peak growth hormone responses to growth hormone releasing hormone were similar to the peak levels obtained following insulin-induced hypoglycemia, glucagon, or L-dopa-propranolol. The height SDS of the patients at the time of study correlated positively with the age at meningitis (r = 0.749, P < 0.05).

One patient had ACTH deficiency with a basal plasma cortisol level of 101 nmol/L (normal level, 150 nmol/L) and a peak level of 318 nmol/L after insulin-induced hypoglycemia. The corresponding levels during corticotropin releasing hormone test were 80 and 413 nmol/L, respectively. For both tests, peak cortisol level was recorded 60 minutes after the injection of secretagogue. Basal plasma ACTH was 5.5 pmol/L (normal range, 2.0 to 11.5 pmol/L). In response to corticotropin releasing hormone, plasma ACTH increased to a peak of 26.2 pmol/L at 60 minutes. The patient had no clinically significant drug history and no history of previous surgery or radiation therapy.

Of the 49 patients, only one had hyperprolactinemia. This patient had a mildly elevated serum prolactin level (28 to 30 g/L) recorded on four separate occasions. The patient had no clinically significant drug history. Direct questioning elicited a complaint of inadequate potency lasting several years before hyperprolactinemia was diagnosed.

Magnetic Resonance Imaging Findings

Scans of the hypothalamicpituitary region revealed abnormal findings in five patients with hypopituitarism (see Table 1). These changes included enhanceable lesions in the hypothalamus, pituitary stalk, or suprasellar cistern (Figure 1); dilatation of the third ventricle due to aqueduct stenosis (Figure 2); adhesion in the basal arachnoid (patient 9) or focal thalamic atrophy (patient 4); and pituitary atrophy of varying severity (see Figure 2). The presence of extensive nodular lesions with abnormal enhancement involving the suprasellar, basal, and interhemispheric cisterns as well as the brain stem (see Figure 1) was found only in patient 2, who was studied 3 years after the meningitic disease.

Correlation of Neurological Presentation with Development of Hypopituitarism

The clinical features at presentation of tuberculous meningitis of the 49 patients and the findings of calcification in their skull radiographs at follow-up are summarized in Table 4. Depending on the level of consciousness, the patients were classified into three clinical stages [13]. Using multivariate stepwise discriminant analysis, only the occurrence of seizure at presentation was identified as a positive predictive factor, with an accuracy of 77.6%, for the development of hypo-pituitarism.

Discussion

In this retrospective study of 49 patients with proven tuberculous meningitis as children, 10 patients were found to have abnormal hypothalamicpituitary function affecting one or more anterior pituitary hormones. Only two patients had previously sought treatment from gynecologists for possible endocrine problems. This apparently high prevalence may be explained in part by an increased readiness among those with endocrine dysfunction to respond to the invitation for investigation. Because this is a nonrandomized study involving only 20% of the study population, it is not possible to conclude from our findings the true prevalence of hypopituitarism following childhood meningitis. Furthermore, our method of recruitment may have resulted in the exclusion of most patients with relatively mild tuberculous meningitis. Despite these limitations, we believe that the prevalence of hypothalamicpituitary dysfunction after recovery from childhood tuberculous meningitis is probably much higher than has been previously recognized, as a result of improved treatment and survival rate of the primary meningitic disease and more regular monitoring with sensitive endocrine investigations.

Growth hormone deficiency was the most common abnormality we observed. The adverse effect on adult height was evident only in those who had tuberculous meningitis before the growth spurt was completed. Although most of the affected patients did not have severely short stature, an earlier diagnosis of growth hormone deficiency and institution of growth hormone therapy would have allowed them to attain a more satisfactory adult height. Because human growth hormone is now widely available, it is important that clinicians caring for children with previous tuberculous meningitis be alerted to this treatable cause of short stature.

In the four patients who had tuberculous meningitis 18 to 36 years previously, growth hormone responses to growth hormone releasing hormone and hypoglycemia (or the other standard pharmacologic tests) were of similar magnitude, suggesting either primary pituitary damage or prolonged deficiency of endogenous growth hormone releasing hormone [14]. On the other hand, in the patient who was studied 3 years after tuberculous meningitis, a large discrepancy was found between the growth hormone responses to growth hormone releasing hormone and hypoglycemia, suggesting a hypothalamic defect in the release of growth hormone releasing hormone [15]. In the same patient, secondary amenorrhea was associated with normal gonadotropin responses to GnRH, suggesting again a hypothalamic defect in GnRH release. The finding of extensive areas of abnormal enhancement in the suprasellar region and normal pituitary structure in her MRI scan further supported a hypothalamic cause for her hypopituitarism. Gonadotropin responses to GnRH also suggested a hypothalamic lesion in two other women with gonadotropin deficiency.

In the only patient with ACTH deficiency, the exaggerated ACTH responses to corticotropin releasing hormone were in keeping with a hypothalamic defect in corticotropin releasing hormone release [11]. We have no proof that his isolated ACTH deficiency occurred as a sequela of childhood tuberculous meningitis, although it is known that in patients with other causes of hypopituitarism, such as cranial irradiation [7], ACTH deficiency can precede the development of other anterior pituitary dysfunction. The alternative diagnosis in this case would be that of idiopathic isolated ACTH deficiency that could be of pituitary or suprapituitary origin [16].

Hyperprolactinemia, another manifestation of suprapituitary dysfunction, was found in only one patient, whereas TSH deficiency was not found at all in this group of patients. We were also unable to find any patient with undiagnosed diabetes insipidus, a well-known sequela of tuberculous meningitis. Presumably most clinicians caring for patients with tuberculous meningitis are aware of this condition that, furthermore, tends to be more symptomatic, so that patients with diabetes insipidus are less likely to have remained undiagnosed.

The areas of abnormal enhancement in the MRI scans of patients 3 and 4 probably represent granulation tissues. With increasing interval from the primary meningitic disease, fibrosis usually occurs so that extensive areas of abnormal enhancement are less likely to be found in patients studied after prolonged intervals. Because we did not do MRI scans in patients with normal endocrine function, it is not known whether some of these patients also had abnormal enhanceable lesions in their hypothalamicpituitary region. It is therefore not possible from this study to establish the pathogenetic significance of these lesions.

This may be analogous to the finding of suprasellar calcification after tuberculous meningitis. Although such calcified deposits are often found in patients with endocrine disturbance, their presence is not invariably accompanied by endocrine disorders [2]. Four of the 39 patients with normal pituitary function in this study had skull radiographs that showed calcified deposits in the vicinity of the sella. Nevertheless, findings from MRI scans in the patients with pituitary dysfunction showed a good correlation with the biochemical tests in that significantly abnormal findings on MRI scans were found only in those with endocrine dysfunction affecting more than one anterior pituitary hormone. They also correlated with the results of endocrine testing and suggested that pituitary dysfunction in these patients occurred as a result of tuberculous lesions affecting the hypothalamus, pituitary stalk, and, directly or indirectly, the pituitary itself.

We conclude that hypothalamicpituitary dysfunction develops in a significant proportion of patients who had tuberculous meningitis in childhood. Because early diagnosis and appropriate replacement therapy can be of great benefit to these patients, clinicians caring for patients who have had tuberculous meningitis should watch for the development of hypopituitarism in their patients. Growth hormone and gonadotropin deficiencies are the most common abnormalities observed, although hyperprolactinemia and ACTH deficiency may also occur. Our study suggests that hypopituitarism is more likely to occur in those who had seizure when their primary meningitic disease occurred. In view of the selection bias in patient recruitment and the retrospective nature of this study, however, further studies are required to identify more reliably the predictive factors for the development of hypopituitarism following childhood tuberculous meningitis.