Postoperative Testing to Predict Recurrent Cushing Disease in Children

Abstract

Context: Postoperative testing after transsphenoidal surgery (TSS) for Cushing disease (CD) in children and its usefulness in predicting residual disease or recurrence are not well studied.

Objective: The objective of the study was to identify which one of three tests that are routinely performed in our institution after TSS performs better in the identification of noncured patients or predict relapse for CD.

Design: This was a retrospective review of clinical data of 72 children who received surgery for CD (age range 5.8–18.3 yr).

Setting: The study was conducted at a tertiary care center.

Methods: After TSS, plasma ACTH and serum cortisol (at 0800 h), urinary free cortisol (UFC) values and an ovine CRH (oCRH) stimulation test were obtained. Patients were followed up for 24–120 months by a formal protocol.

Results: Of 72 children with CD, 66 (94%) achieved sustained remission after TSS. Two children had persistent disease after TSS, whereas four children appeared cured at first but relapsed later. All four had low or undetectable UFCs that were not different from cured patients (P > 0.0.1). Children who remained in remission had significantly lower morning ACTH and cortisol levels after TSS compared with those who relapsed (P < 0.001). During an oCRH stimulation test, ACTH and cortisol values were higher in patients who relapsed vs. those in remission (P <0.001). Lack of histological confirmation of an adenoma, normal serum cortisol or ACTH, a normal response to oCRH, and glucocorticoid replacement for less than 6 months after surgery were associated with relapse.

Conclusion: In pediatric patients with CD, low UFCs after TSS are not good predictors of sustained remission; morning ACTH and cortisol values and/or an oCRH test after TSS predicted patients that recurred.

After transsphenoidal surgery for Cushing’s disease, the postoperative ovine CRH stimulation test, morning ACTH, and cortisol are useful tests in identifying pediatric patients with persistent disease, or at risk of recurrence.

A pituitary ACTH-secreting adenoma [Cushing disease (CD)] accounts for more than 75% of the cases of Cushing syndrome (CS) in children (1). Transsphenoidal surgery (TSS) is the treatment of choice (1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16). Remission of CS in children after TSS varies widely from center to center: cure rates as low as 45% and as high as 97% have been reported (1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16). Longitudinal studies of pediatric patients with CD showed that hypercortisolemia may recur, sometimes as late as in the second decade after an apparently successful TSS; however, most patients with CD relapse present within the first 2 yr after surgery (1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16). Consequently, testing in the immediate postoperative period that assesses cure and can be used to predict recurrence is essential for optimal management of these children. Several diagnostic modalities have been used in studies of adult patients with CD after TSS (17,18,19,20,21,22,23): measurement of serum cortisol or ACTH, determination of urinary free cortisol (UFC) excretion, and low-dose dexamethasone suppression test have all been used with varying success (17,18,19,20,21,22).

In addition to the above tests, at the National Institutes of Health (NIH), we have used the ovine CRH (oCRH) stimulation test after TSS according to the protocol proposed by Avgerinos et al. (23): a normal response to oCRH indicated the presence of residual pituitary tumor in that study. Recent series of adult patients with CD confirmed a higher recurrence rate in patients with normal oCRH testing in the immediate postoperative period (23,24,25,26,27).

Only one study has addressed the role of the oCRH test after TSS in predicting recurrence in pediatric patients with CD (11); only three children were studied: one child had subnormal ACTH and serum cortisol response to oCRH administration and remained in remission; the second child had subnormal serum cortisol response but an excessive ACTH response to oCRH and his CD recurred 3.2 yr later. The third child, who had persistent disease, also had high responses of both cortisol and ACTH to oCRH (11).

In the present study, we studied our large cohort of children that we have treated for CD over the last 10 yr. Only patients with at least 2 yr of follow-up data and only those who came back to the NIH for at least one visit were included. There were only a small number of patients with CD who were not cured in this cohort; nevertheless, the data point to useful conclusions about what constitutes remission of pediatric CD in the immediate post-TSS period and which tests identify children with persistent or disease relapse in the first few years after surgery.

Subjects and Methods

Subjects and protocol

Seventy-two children who presented with classic stigmata of CD (Table 1) and were admitted to the NIH Warren Magnuson Clinical Center in the last 10 yr were included in the analysis. The studies were under clinical protocol 97-CH0076 that was approved by the Eunice Kennedy Shriver National Institute of Child Health and Human Development Institutional Review Board. Informed consent from the patients’ parents (and assent from older children) was obtained for all patients. A general endocrine screening protocol was used for non-CD-related testing (28). Some of these children have been previously published and the diagnosis of CD was confirmed, as previously described (29,30,31). All children underwent TSS at the NIH.

Table 1.

Frequency of clinical findings in children with Cushing’s disease

Physical findings	No. of patients	%
Obesity (body mass index Z-score >2.0)	45	62
Growth retardation	55/67a	82
Short stature	28/67a	42
Irregular menses	14/19	73
Hypertension	33b	46
Hirsutism	14/36c	39
Hyperpigmented abdominal striae	29	40
Acne	33	46
Fatigue	30	41
Early secondary sexual characteristics	3/8d	34
Delayed puberty	2/26e	7
Bruising	16	22
Mood changes	20	28
Muscle weakness	9	13
Achantosis nigricans	19	26
Sleep disturbances	8	11
Headaches	24	33

^aChildren with pubertal Tanner stage I–IV. 

^bHypertension was defined as systolic or diastolic blood pressure above 2 sd for age and sex. 

^cSample was composed of females only. 

^dSample composed of girls younger than 8 yr old and boys younger then 9 yr old. 

^eSample included girls older than 13 yr old and boys older than 14 yr of age. 

Postoperative evaluation

There were several evaluations done including the following: 1) morning serum cortisol (0800 h), in which patients received glucocorticoid coverage during the first 36 h of the perioperative period, and steroids were weaned off postoperatively and serum cortisol measurements were taken for 3 consecutive days, starting after the third postoperative day (POD) while the patient was on no replacement (all children were monitored and none experienced adrenal insufficiency; replacement glucocorticoids were started on the fifth POD at a dose of 0.5 mg/d dexamethasone, adjusted for weight); 2) morning (0800 h) serum ACTH; 3) 24-h UFCs were measured also after the third POD and were continued until the day before the oCRH test; and 4) oCRH stimulation test: all except one child (71 of 72) underwent oCRH testing after surgery. The oCRH stimulation test was performed on the 10th POD as previously described (23). A venous catheter was placed in the forearm 1 h before testing and oCRH was administered (1 μg/kg of body weight). Samples for cortisol and ACTH were taken 15 min before and just before and 15, 30, 60, 90, and 120 min after oCRH. Steroids were withheld at least 24 h before the oCRH test (23).

Remission of CD

Children were considered in remission in the immediate postoperative period if average UFCs were less than 20.0 μg/m² · d or less than 55.2 nmol · d (normal 20.0–70.0 μg/m² · d or 55.2–193.0 nmol · d) (23,29,30,31). After their discharge from the hospital, all patients were started on glucocorticoid replacement (hydrocortisone) (at 8–12 mg/m² · d).

Long-term follow-up

Our final sample was composed of 72 of 79 children for whom at least 24 months have passed after TSS, and we have data for: seven children, however, were excluded from the present study because two did not complete testing; another had a seizure and did not undergo any testing during the immediate postoperative period; and four children had no further follow-up at the NIH after their TSS. The first three of these seven children remain in remission to date according to outside doctor reports.

From the 72 children included in the study, two children had persistent disease, 44 returned during the first year after TSS, whereas the remaining 26 returned at different intervals and after they had discontinued glucocorticoid replacement. Recovery of normal adrenocortical function was evaluated by an ACTH stimulation test that was obtained only if clinically indicated if the patient came back after the first 12 months (data not shown).

Duration of follow-up for these children ranged from at least 24, up to 120 months. These data included the following: 1) clinical stigmata of CS; 2) biochemical testing (UFC, cortisol circadian rhythm); 3) current endocrine medications; and 4) requirement of additional treatment for CD.

Recurrence or relapse of CD

Recurrence or relapse was documented in four patients whose midnight serum cortisol levels exceeded 4.4 μg/dl (121.0 nmol/liter) and UFCs were above 70.0 μg/m² · d (193.0 nmol/d) (29,30,31). All hormone assays were performed, as previously described (29,30,31).

Statistical analysis

All continuous and descriptive statistics data are expressed as mean ± sd. For comparison groups data are expressed as mean ± sem. Categorical variables were expressed as frequencies and compared using χ² and Fisher’s exact test when appropriate. Measurements of ANOVA were performed for numerical data for the comparison of multiple sample means. The Wilcoxon rank-sum test was used to analyze numeric variables. Receiver operating characteristic curves and 95% confidence intervals were determined for sensitivity and specificity values of all tests, as previously described (29,30,31). The best balance between sensitivity and the specificity are presented (supplemental Tables S1 and S2, published as supplemental data on The Endocrine Society’s Journals Online Web site at http://jcem.endojournals.org). Clinical and biochemical predictors of remission or disease relapse were evaluated by the construction of a forward stepwise regression model. Variables analyzed in the model as independent predictors of disease relapse were age, sex, mean UFC before surgery, mean ACTH before surgery, preoperative oCRH stimulation test, magnetic resonance imaging (MRI) results, dural invasion, bilateral inferior petrosal sinus sampling, pathology findings, and surgical findings. The pre- and post-TSS biochemical testing (ACTH, UFC, and serum cortisol) were transformed to a logarithmic scale to achieve approximately normal distribution. P < 0.10 was used to enter the model and P > 0.15 to leave the model. For all statistical comparisons, P < 0.05 was considered significant. Data were analyzed using the Stata statistical software (release 9; Stata Corp. LP, College Station, TX).

Results

Patient description and clinical course

Persistent disease after TSS

Seventy-two consecutive children who presented with typical CS were studied (Tables 1 and 2). Mean duration of clinical symptoms (from onset of symptoms to diagnosis) was 28 months (range 3–84 months). Four had undergone a first TSS elsewhere. There were two children with persistent hypercortisolism after their TSS at the NIH: a 10-yr-old boy and a 16-yr-old girl. The former had a TSS elsewhere before coming to the NIH; he had a normal pituitary MRI and no tumor was found at surgery at either operation; hypercortisolism persisted despite excision of 75% of his anterior lobe. The other patient had an invasive ACTH-secreting macroadenoma. Both patients were placed on ketoconazole and received radiation therapy later. Thus, immediate remission after TSS occurred in 97% (70 of 72) of our cohort.

Table 2.

Demographic data and diagnostic features and procedures

	Persistent (n = 2)	Remission (n = 66)	Relapsed (n = 4)
Age	10, 16a	13 ± 3	11.1 ± 4
Sex (female/male)	1/1a	34/32	1/3
Body mass index Z-score	1.9, 2a	1.9 ± 0.7	2.2 ± 0.6
Height age Z-score	−3, −1a	−0.9 ± 1.5	−1.4 ± 0.9
UFC (normal 20–70 μg/m2 · d)	274; 2052a	202 ± 31	54 ± 82
Midnight serum cortisol (normal <4.4 μg/dl)	29, 136a	20 ± 10	14 ± 2
oCRH stimulation test
Cortisol increase >20%	100%	75%	75%
Percent increase	112%, 19%a	108% ± 114%	58% ± 56%
Corticotropin increase >35%	100%	83%	100%
Percent increase	245%, 24%a	240% ± 162%	130% ± 61%
HDDST cortisol suppression >20%	50%	100%	100%
Percent decrease	85%, 15%a	75% ± 22%	86% ± 2%
MRI findings (visualization of an adenoma)	50%	67%	25%
Underwent BIPSS	50%	64%	50%
Pathology confirmation	50%	85%b	33%b

Data are presented as mean ± sd. BIPSS, Bilateral inferior petrosal sinus sampling; HDDST, high-dose dexamethasone suppression test. To convert UFC (micrograms per day) to System International units (nanomoles per day), multiply by 2.759. To convert midnight cortisol (micrograms per deciliter) to System International units (nanomoles per liter), multiply by 27.59. To convert morning corticotropin (picograms per milliliter) to System International units (picomoles per liter), multiply by 0.2202. 

^aChild with macroadenoma. 

^bIn those whom a tumor was found, statistically significant P < 0.05. 

Initial remission and recurrence later

Four children who initially achieved remission relapsed 5–30 months after TSS. In two patients with negative preoperative MRI, postoperative UFCs dropped to less than 1.0 μg/m² · d (2.8 nmol/d). In both cases, an encapsulated microadenoma was removed at surgery, but histologically the tissue did not stain for ACTH. Hypercortisolism recurred 5 and 18 months later, respectively. In each patient, an encapsulated ACTH-stained adenoma was found and removed from the opposite side of the gland at repeat TSS. Although both patients showed hypocortisolism after their second TSS, one was eventually referred for radiation therapy because infiltration of the cavernous sinus by the tumor was noted during his second TSS and his biochemical testing remained consistent with CD. In the other patient, no tumor could be identified during the second TSS and hemihypophysectomy was performed. Again, no adenoma was seen histologically, and despite immediate post-second TSS hypocortisolism, he was found to have recurrent CD 27 months later. Another patient with a positive pre-TSS MRI had tumor invasion of the cavernous sinus at surgery. He was hypocortisolemic for only 10 months. Thus, sustained remission after an initial TSS at the NIH was achieved in 92% (66 of 72) of children. Of these 66 children with sustained remission, 95.5% (63 of 66) had no previous surgery for CD. For these patients, median age was 13 yr (range 5.8–18.3; mean 13 ± 3). There were 34 females and 32 males, whereas from the four patients with recurrent disease, three were males with a median age of 11.2 yr (range 7.2–14.9; mean 11.1 ± 4). Children with recurrent disease had no previous surgery for CD. Clinical and biochemical characteristics of these patients are presented in Tables 1 and 2.

Testing in the immediate postoperative period

Morning Plasma ACTH

Overall, there was a downward trend in morning ACTH levels after TSS (Fig. 1A). The mean of two morning ACTH values was significantly higher in children who had persistent disease [at 34.0 pg/ml (7.5 pmol/liter) and 67.0 pg/ml (15.0 pmol/liter), respectively], compared with those who went into remission. Mean ACTH for patients who relapsed later was 21.0 ± 10.0 pg/ml (5.0 ± 2.2 pmol/liter) in the immediate post-TSS time vs. 5.3 ± 0.5 pg/ml (1.2 ± 0.1 pmol/liter) in those who remain in remission to date (P < 0.001). A mean morning serum ACTH above 17.0 pg/ml (4.0 pmol/liter) identified three of four children who later recurred with a sensitivity of 75% [three of four (95% confidence interval (CI₉₅) 20–96%)]. A value less than 17.0 pg/ml (4.0 pmol/liter) identified 64 of 65 children (one child did not have an ACTH measurement) who remain in remission to date, and the test specificity was 98.5% [64 of 65, (CI₉₅ 92–100%)]. A peak morning plasma ACTH of 17.5 pg/ml (4.0 pmol/liter) identified three of four children who later recurred, with a sensitivity of 75% [three of four (CI₉₅ 20–96%)] and specificity of 98.5% [64 of 65 (CI₉₅ 92–100%) Fig. 1B]. A morning plasma ACTH nadir value less than 10.8 pg/ml (2.4 pmol/liter) identified 63 of 65 of the children who remained in remission, and the test specificity was 97% [63 of 65 (CI₉₅ 89–100%)], whereas a plasma ACTH nadir above 10.8 pg/ml (2.4 pmol/liter) was observed in three of four children with recurrent disease [sensitivity of 75%, (CI₉₅ 20–96%)] (Fig 1C). Supplemental Table S1 shows the sensitivity and specificity of the plasma ACTH nadir values to indicate remission. As in adult patients with CD, remission or cure of CD was defined in our series by the presence of adrenal insufficiency; a plasma ACTH nadir was used for the rest of the calculations (6,19,20,21,22,32,33,34).

Figure 1.

Morning corticotropin (ACTH) after TSS. A, There was a significant decrease in mean plasma ACTH before and after TSS. B, Peak plasma ACTH. C, Plasma ACTH nadir. D, There was a significant decrease in mean morning serum cortisol before and after TSS. E, Peak serum cortisol. F, Serum cortisol nadir. Mean and sem are shown when sem is greater than the data point. *, Significant P < 0.05. The y-axis represents hormonal values after surgery; the x-axis represents the different study groups. NS, Not significant.

Morning serum cortisol

Overall, there was a decrease in morning serum cortisol after surgery in all groups (Fig. 1D). The two children with persistent disease had mean morning serum cortisol values of 19.0 μg/dl (524.2 nmol/liter) and 35.0 μg/dl (965.7 nmol/liter), respectively. Mean post-TSS cortisol values (from the third, fourth, and fifth POD) of children who later recurred were overall higher than those that remain in remission (5.0 ± 3.0 μg/dl or 138.0 ± 83.0 nmol/liter vs. 2.0 ± 0.13 μg/dl or 55.2 ± 3.6 nmol/liter; P < 0.001). A mean morning serum cortisol of 6.5 μg/dl (179.3 nmol/liter) provided the most specific value to differentiate between remission and recurrent disease, sensitivity of 50% [two of four (CI₉₅ 8–92%)]. All 66 children who remain in remission to date had a mean morning cortisol value after TSS less than 6.5 μg/dl (179.3 nmol/liter), specificity 100% [66 of 66 (CI₉₅ 95–100%)]. A peak serum cortisol of 16.0 μg/dl (441.4 nmol/liter) provided a sensitivity of 25% [one of four (CI₉₅ 4–80%)] and specificity of 100% [66 of 66 (CI₉₅ 95–100%)] (Fig. 1E). A single serum cortisol nadir less than 4.7 μg/dl (130.0 nmol/liter) after TSS provided the most accurate value to differentiate between remission and recurrent disease with a sensitivity of 50% [two of four (CI₉₅ 8–92%)] and specificity of 100% [66 of 66 (CI₉₅ 95–100%)] (Fig. 1F). Supplemental Table S2 shows the sensitivity and specificity of the serum cortisol nadir to indicate remission; again, as in adult patients, remission or cure of CD was defined by the presence of adrenal insufficiency and a serum cortisol nadir was used for the rest of the calculations (6,19,20,21,22,32,33,34).

UFC

UFC excretion decreased after TSS significantly in all patients (P < 0.001) (Fig. 2). Hypoadrenalism was documented in all four patients that later recurred. In these four patients, low (<20.0 μg/m² · d or 55.2 nmol/d) or undetectable UFC levels in the immediate postoperative period were documented.

Figure 2.

UFC after TSS (A), UFC trend (B), peak UFC and UFC nadir (C) are shown.

Postoperative oCRH stimulation

During the postoperative oCRH stimulation test, baseline plasma ACTH levels and the integrated and peak responses were significantly higher at all times in patients who experienced recurrence, compared with those who went into remission (P < 0.001) (Fig. 3A). Basal and peak-stimulated plasma ACTH after oCRH administration are shown in Fig. 3, B and C, respectively. A peak-stimulated plasma ACTH above 32.0 pg/ml (7.0 pmol/liter), after oCRH administration, was observed in all children who recurred, providing a sensitivity of 100% [four of four (CI₉₅ 31–100%)]. Nevertheless, five children who underwent sustained remission had a peak-stimulated ACTH greater than 32.0pg/ml (7.0 pmol/liter), giving a specificity of 92% [61 of 66 (CI₉₅ 83–97%)].

Figure 3.

Results of the postoperative oCRH stimulation test. During the oCRH test, mean plasma ACTH (A) was significantly higher in patients who recurred than in those who underwent remission at all time points. The y-axis represents plasma ACTH; the x-axis represents time in minutes of corticotropin collection. Basal (B) and peak-stimulated ACTH levels (C) in response to oCRH administration for the patients in the study [y-axis represents plasma ACTH (picograms per milliliter) and x-axis represents the different study groups]. D, Serum cortisol was significantly higher during the oCRH test in patients with recurrent disease than in those who underwent remission at all time points. The y-axis represents serum cortisol (micrograms per deciliter); the x-axis represents time in minutes of cortisol collection. Basal (E) and peak-stimulated serum cortisol levels (F) in response to oCRH administration for the patients in the study are shown [y-axis represents serum cortisol (micrograms per deciliter) and x-axis represents the different study groups]. Mean and sem are shown when sem is greater than the data. In panels A and D, a triangle (▴) represents children in remission and children with recurrent disease are represented as a square (▪). A circle (○) represents a patient with persistent disease.

Also, during the postoperative oCRH stimulation test, serum cortisol mean baseline levels and integrated and peak response were significantly higher at all times in patients who recurred compared with those who went into remission (P < 0.001) (Fig. 3D). Basal and peak-stimulated cortisol after oCRH are shown in Fig. 3, E and F, respectively. A peak-stimulated serum cortisol greater than 10.0 μg/dl (276.0 nmol/liter) after oCRH identified all children with recurrent disease with a sensitivity of 100% [four of four (CI₉₅ 31–100%)]. However, eight children with sustained remission also had a normal serum cortisol response to oCRH, giving the test a specificity of 88% [58 of 66 (CI₉₅ 76–94%)].

Combining both peak cortisol and peak ACTH responses to oCRH administration identified all children with recurrent CD and increased the test specificity to 95% [63 of 66 (CI₉₅ 86–99%)]. The oCRH test also was performed in one of two children with persistent disease: a marked response of serum cortisol and ACTH after oCRH administration was observed in this child, as has been described by others (11).

False-negative and false-positive results

No false-negative results were reported for the postoperative oCRH stimulation test. False-negative rate for the morning corticotropin and morning serum cortisol was low at 1.6% (one of 64) and 3% (two of 68), respectively. No false-positive tests were reported for the morning serum cortisol. In contrast, both the morning corticotropin and the postoperative oCRH stimulation test had a high false-positive rate at 40% (two of five) and 43% (three of seven), respectively.

Test predictive values

The test with the highest positive predicted value was serum cortisol at 0800 h: the probability of recurrence for a child with serum cortisol above 4.7 μg/dl (130.0 nmol/liter) was 100% [two of two (CI₉₅ 20–100%)]. The positive predicted value for the postoperative oCRH stimulation test was 57% [four of seven (CI₉₅ 20–88%)] and for morning ACTH 60% [three of five (CI₉₅ 17–93%)]. In contrast, the test with the highest negative predictive value was the oCRH stimulation test: the probability of a child with a subnormal oCRH response remaining in remission was 100% [63 of 63 (CI₉₅ 93–100%)]. The negative predictive value for morning serum ACTH was also high at 98% [63 of 64 (CI₉₅ 91–100%)] and for morning serum cortisol was high as well at 97% [66 of 68 (CI₉₅ 89–99%)].

Diagnostic utility of each test

The diagnostic accuracy of a morning serum cortisol nadir (0.75) was lower than the post-TSS oCRH stimulated values of cortisol (0.96) and ACTH (0.97) as well as spot morning ACTH (0.99).

Factors associated with increased risk of recurrent disease

None of the clinical and biochemical parameters before surgery were independent predictors of recurrence for CD in these children. Post-operatively, as expected, we found that positive and independent predictors of recurrent CD were lack of histological confirmation of CD (r = 0.10; P = 0.048), a normal response to post-TSS oCRH stimulation (peak cortisol > 10.0 μg/dl or 276.0 nmol/liter and peak-stimulated ACTH > 32.0 pg/ml or 7.0 pmol/liter; r = 0.24; P < 0.001), and, finally, a normal morning serum ACTH (>10.8 pg/ml or 2.4 pmol/liter; r = 0.28; P < 0.001) and cortisol (>4.7 μg/dl or 130.0 nmol/liter; r = 0.44; P < 0.001).

Forty-four patients underwent testing of their adrenocortical function during their first year after surgery in our center. In these children, glucocorticoid replacement therapy that was needed for only 6 months or less was associated with recurrent disease (r = 0.90; P < 0.001).

Discussion

After surgical resection for CD, children can experience persistent disease, long-term remission, or recurrence (1) (Table 3). The majority of relapses occur within the first few years after surgery (1,2,3,4,5,6,8,9,10,11,12,13,15). Because CD is a debilitating condition that significantly affects growth and development of these children, tests that could predict long-term remission or recurrence are of importance in their management. Our study suggests that the post-TSS oCRH stimulation test, morning ACTH, and cortisol are useful tests in identifying pediatric patients with persistent CD or at risk for recurrent CD.

Table 3.

Time to relapse after TSS in pediatric patients with Cushing’s disease

Reference	n	Age (yr)	Remission	Persistent	Recurrence or relapse	Follow-up (months)	Time to recurrence (months)
Styne et al. 1984 (5)	15	10 [7–13]	93% (14/15)	7% (1/15)	7% (1/14)	44 [10–109]	48
Magiakou et al. 1994 (1)	49	<21	98% (48/49)	2% (1/49)	6% (3/48)	22 [5–60]	4, 14, and 24
Dyer et al. 1994 (12)	33	<16	70% (23/33)	30% (10/33)	10% (2/21)a	[9–168]	Over 36
Partington et al. 1994 (3)	16b	<17	80% (12/15)c	20% (3/15)	25% (3/12)	54 [4–163]	98, 123, and 63
Leinung et al. 1995 (8)	22	15.5 [10–18.9]	77% (17/22)	23% (5/22)	41% (7/17)	80.4 [15.6–180]	84 [7.2–150]
Mindermann and Wilson 1994 (9)	42	<20	83% (35/42)	17% (7/42)	17% (6/35)	49	N/A
Knappe and Ludecke 1996 (10)	55	14 [4.2–18.9]	96% (53/55)	4% (2/55)	17% (9/53)	54 [2–184]	3–60
Devoe et al. 1997 (11)	42	13.1 [6.5–18]	83% (35/42)	17% (7/42)	27% (7/26)d	86 [18–163]d	50 [9–74]
Massoud et al. 1997 (13)	14e	14 [8.7–16.3]	85% (11/13)f	15% (2/13)	18% (2/11)	84 [12–168]	10 and 12
Joshi et al. 2005 (14)	25	13.4 [6.6–17.8]	60% (15/25)	40% (10/25)	N/A	59.5 [6–128]	N/A
Kanter et al. 2005 (4)	33	13 [5–19]	76% (25/33)	24% (8/33)	12% (3/25)	44 [2–108]	18, 81, and 92
Das et al. 2007 (15)	10	15 [12–17]	70% (7/10)	30% (3/10)	43% (3/7)	82 [24–120]	36, 60, and 92
Hoffman et al. 2008 (16)	38	<18	86% (31/36)g	14% (5/36)	N/A	At least for 3 months	N/A
Present study	72	13.3 [5.8–18.3]	97% (70/72)	3% (2/72)	6% (4/70)	30 [24–120]	5, 10, 18, and 27

The mean is presented with the range in brackets. N/A, Not applicable. 

^aTwo of 23 patients had no follow-up. 

^bIncluded two patients who had Nelson syndrome (n = 16). 

^cExcluded one patient with no follow-up data (n = 15). 

^dFollow-up was available for 26 of 42 after 18 months. 

^eIncluded one patient with Neslon syndrome (n = 14). 

^fExcluded one patient with Nelson syndrome (n = 13). 

^gPatients with selective adenectomy. 

It has already been suggested (in adult patients with CD) that a normal response to oCRH in the immediate postoperative period may indicate residual tumor (23,24,25,26,27). To our knowledge, only one published study examined oCRH-induced corticotropin and cortisol secretion in children with CD after TSS, and there were only three children in that report (11). In the present series, 71 of 72 children with CD underwent an oCRH stimulation test after TSS; this included 66 children with sustained remission, four with recurrent disease, and one child with persistent disease. Our cohort, despite low number of recurrences, still represents the largest series of pediatric patients with CD in whom extensive testing (including oCRH stimulation) was performed after TSS.

In our study, recurrence or CD relapse was observed in 57% of children who during the early postoperative period had a normal response to oCRH; in adult patients, recurrence was reported in 23–50% of patients who had a normal response to oCRH in the early postoperative period (23,24,26). Accordingly, no delayed relapse was observed among children who after surgery did not respond to oCRH administration; the probability of a child with a subnormal oCRH response remaining in remission at 24 months was 100%. Thus, in children, as in adults, initial and sustained remission is more likely in patients with subnormal response to oCRH in the immediate postoperative period (23,24,25,26).

Our study is also the first to report on the value of morning ACTH levels to predict remission or recurrent disease after surgery in children with CD. As Fahlbusch et al. (6), who studied adult patients with CD, we found that a subnormal ACTH level of less than 10.8 pg/ml (2.4 pmol/liter) was the best cutoff value for differentiating between remission and relapse. In accordance, the majority of children who remained in remission had a subnormal ACTH level after surgery, as described in series of mainly adult patients (6).

Several studies in children have reported the value of morning cortisol in predicting remission or recurrent disease (2,4,8,10,11,15). Knappe and Ludecke (10) found that, after TSS for CD in children, those who had a mean serum cortisol less than 6.0 μg/dl (166.0 nmol/liter) remained in remission in 97% (31 of 32) of the cases. Recurrence rate was 42% (five of 12) in children whose cortisol levels were between 6.0 and 25.0 μg/dl (166.0 and 690.0 nmol/liter) (10). Of particular importance is the fact that glucocorticoids were withheld safely perioperatively and postoperatively, while morning serum cortisol levels were sampled (10); glucocorticoid replacement was started after establishing hypocortisolemia on postoperative d 1 (10). In agreement with Knappe and Ludecke, we found that patients who underwent remission had a mean serum cortisol value less than 6.5 μg/dl (179.3 nmol/liter). In addition, we found that an 0800 h serum cortisol nadir of 4.7 μg/dl (130.0 nmol/liter) was the best cutoff value to establish remission. This value is close to the serum cortisol cutoff value used to establish remission after TSS for CD in adult patients, which is 5.0 μg/dl (138.0 nmol/liter) (22,34).

Other serum cortisol cutoff values that have been used to establish remission after surgery have included 1.8 μg/d (50.0 nmol/liter) (32). Using a 1.8 μg/dl (50.0 nmol/liter) for serum cortisol would have identified 50% of children who relapsed in our study and 88% of those who underwent remission. However, 12% of these children would have undergone an apparently unnecessary second intervention. In contrast, using the 5.0 μg/dl (138.0 nmol/liter) for serum cortisol provided similar sensitivity (50%) but a higher specificity (100%); its low sensitivity can be overcome by the concurrent use of another test, namely the oCRH test (22,34).

A test that appears to be of limited use in predicting remission or recurrence or CD in the immediate postoperative period is urinary steroid excretion. Low and even undetectable UFC levels were found in all children in our series, whether they underwent sustained remission or whether they recurred. Of particular interest were two children who did not achieve sustained remission; these two children also had subnormal serum cortisol levels, low UFCs, and adrenal insufficiency in the immediate postoperative period. The lowest serum cortisol documented after surgery was less than 1.0 μg/d (<28.0 nmol/liter). These cortisol levels were similar to children who achieved and sustained remission. However, mean ACTH levels in these two children were normal and they both had a normal oCRH response after surgery. These findings suggest that patients with a normal response to postoperative oCRH stimulation who also have a normal morning serum ACTH [despite morning serum cortisol levels less than 4.7 μg/dl (130.0 nmol/liter) or low/undetectable UFC levels] are more likely to experience recurrent disease. This was confirmed by the receiver operating characteristic curve, which showed that combined use of postoperative oCRH test and morning ACTH levels are superior to the morning serum cortisol for distinguishing remission from recurrent disease.

Our findings suggest that dynamic testing might be a better indicator of sustained remission, and they are in agreement with those by Valero et al. (36) and Carrasco et al. (35) in adults with CD. Valero et al. (36) found that a desmopressin stimulation test after TSS was a better predictor of disease relapse than early postoperative cortisol measurements, whereas Carrasco et al. (35) found that midnight salivary cortisol was a better indicator of remission than UFC.

Studies in adult patients have suggested several factors associated with increased risk for CD relapse: younger age at diagnosis, longer duration of symptoms, severe clinical picture, depression or behavioral symptoms, significantly elevated serum cortisol and UFC levels and exaggerated responses to oCRH stimulation preoperatively, lack of visualization of an adenoma by MRI, evidence of dural or sinus invasion at surgery, normal response to postoperative oCRH, normal ACTH and cortisol levels postoperatively, and short duration of glucocorticoid replacement (6,16,17,26,35,36,37,38,39). In contrast, factors associated with increased remission rate include identification of a tumor at operation and histologic confirmation of an adenoma (33,35,39). Indeed, in our study, too, lack of histological confirmation for an ACTH-producing tumor for CD and a need for glucocorticoid replacement therapy for less than 6 months after surgery both predicted recurrence of CD along with the normal postoperative biochemical testing.

Some of the limitations of our study included the retrospective nature of collected data, the small number of children in the recurrent or relapsed CD group, and that not all children returned at their scheduled visit (and, thus, some of the testing was not performed in our center). Nevertheless, our findings are similar to adult patients with pituitary surgery for CD (6,23,24,25,26,27,34,35). It will be very important to document follow-up of these children for another 5–10 yr because late recurrences are known to occur (3,4,5,8,10,11,12,15).