Outcome of Surgical Treatment of 200 Children With Cushing's Disease

Abstract

Context:

Factors influencing the outcome of surgical treatment of pediatric Cushing's disease (CD) have not been fully established.

Objective:

The aim of this study was to examine features influencing the outcome of surgery for pediatric CD.

Design:

In this prospective observational study, the clinical, imaging, endocrinological, and operative outcomes were analyzed in consecutive patients treated at the National Institutes of Health (NIH) from 1982 through 2010.

Setting:

The study was conducted in a tertiary referral center.

Results:

Two hundred CD patients (106 females, 94 males) were included. Mean age at symptom development was 10.6 ± 3.6 years (range, 4.0 to 19.0 y). Mean age at NIH operation was 13.7 ± 3.7 years. Twenty-seven patients (13%) had prior surgery at another institution. Magnetic resonance imaging identified adenomas in 97 patients (50%). When positive, magnetic resonance imaging accurately defined a discrete adenoma in 96 of the 97 patients (99%), which was more accurate than the use of ACTH ratios during inferior petrosal sinus sampling to determine adenoma lateralization (accurate in 72% of patients without prior surgery). A total of 195 of the 200 patients (98%) achieved remission after surgery (189 [97%] were hypocortisolemic; 6 [3%] were eucortisolemic postoperatively). Factors associated with initial remission (P < .05) included identification of an adenoma at surgery, immunohistochemical ACTH-producing adenoma, and noninvasive ACTH adenoma. Younger age, smaller adenoma, and absence of cavernous sinus wall or other dural invasion were associated with long-term remission (P < .05). A minimum morning serum cortisol of less than 1 μg/dl after surgery had a positive predictive value for lasting remission of 96%.

Conclusions:

With rare disorders, such as pediatric CD, enhanced outcomes are obtained by evaluation and treatment at centers with substantial experience. Resection of pituitary adenomas in pediatric CD in that setting can be safe, effective, and durable. Early postoperative endocrine testing predicts lasting remission. Because lasting remission is associated with younger age at surgery, smaller adenomas, and lack of dural invasion, early diagnosis should improve surgical outcome.

Excessive ACTH excretion by a pituitary adenoma is the cause of Cushing's disease (CD). In children, CD often presents with obesity, hypertension, delayed puberty, and cessation of linear growth (1, 2). Because adenoma resection can preserve pituitary function and is immediately effective, adenoma removal is the preferred treatment for CD. Correlation of the results of preoperative endocrine assessment, pituitary imaging, surgical findings, factors that affect surgical success, or the accuracy of the results of postoperative endocrine assessment to predict lasting remission has not been established. We analyzed a large, consecutive cohort of pediatric CD patients evaluated and treated prospectively at a single institution.

Patients and Methods

Patient inclusion

Consecutive pediatric CD patients (age < 21 y at time of surgery) who underwent endocrinological evaluation and pituitary surgery at the National Institutes of Health (NIH) from 1982 through 2010 were included. Data for 49 of the patients have been previously reported (1). Patients and/or their guardians gave consent to participate in Institutional Review Board-approved protocols.

Clinical evaluation

In this prospective observational study, pre- and postoperative evaluations were performed. Telephone, e-mail, and/or mail contact supplemented data from clinical evaluations, as in prior studies from the NIH (3, 4).

Endocrine evaluation

Preoperative evaluation

Testing for excess cortisol production (Cushing's syndrome) and the underlying cause of hypercortisolism was performed as previously described (1, 5). After completing a study to determine the diagnostic utility of inferior petrosal sinus sampling (IPSS) (6), only patients that had negative preoperative magnetic resonance imaging (MRI) or an indefinite endocrine evaluation underwent bilateral IPSS to determine the source of ACTH secretion (6–8).

Postoperative evaluation

Studies to assess biochemical remission after surgery were performed on postoperative days 3 to 5, as previously described (3, 4). Patients received dexamethasone .5 mg, given iv or orally every 6 hours for 5 or 6 doses, beginning on the day of surgery. Some had a sixth dose of 1 mg given at 11 pm on the first postoperative day. Serum cortisol and urinary free cortisol (UFC) were obtained daily from postoperative day 3, until hypocortisolism (cortisol < 5 μg/dl [138 nmol/L] and UFC less than the lower limit of normal [see below]) was achieved in at least 1 sample (3, 4).

Biochemical remission was categorized as “hypocortisolism” if the morning serum cortisol nadir was less than or equal to 5 μg/dl or a 24-hour UFC nadir of less than the lower limit of the contemporary reference range. Biochemical remission was categorized as “eucortisolism” if the morning cortisol and the 24-hour UFC were within the reference range (4).

Hormone assays

The 24-hour UFC excretion was measured by RIA (SmithKline Beecham, Van Nuys, California) from 1982 to September 1994, with a reference range of 20–90 μg/d; by RIA (Mayo Medical Labs, Rochester, Minnesota) from September 1994 to October 2001, with a reference range of 24–108 μg/d; by chemiluminescence immunometric assay (Nichols Advantage; Nichols Institute Diagnostics, Inc, San Juan Capistrano, California) from October 2001 to December 2005, with a reference range of 8–77 μg/d; and by liquid chromatography, tandem mass spectrometry (Mayo Medical Labs, Rochester, Minnesota; or the NIH Department of Laboratory Medicine) from December 2005 to the present, with a reference range varying by age (3–9 y, 1.4–20 μg/d; 9–13 y, 2.6–37 μg/d; 13–18 y, 4.0–56 μg/d; and greater than 18 y, 3.5–45 μg/d).

Serum cortisol was measured by RIA until 1989 and then by immunoassay until the present, with morning reference ranges of 5–7 to 25–26 μg/dL (4).

Imaging evaluation

From 1982 to 1986, contrast-enhanced computed tomography was used for pituitary imaging. From 1983 to 1997, pituitary MRI with standard spin echo sequences, and from 1997 to 2010, standard spin echo sequences and spoiled gradient recalled (SPGR) acquisition sequences were used (9).

Adenoma removal

The pituitary gland was accessed via a sublabial, transsphenoidal approach using the operating microscope. For adenomas identified at surgery, microsurgical dissection around the histological pseudocapsule encasing the tumor was used for removal (10). For adenomas not visualized after thorough examination of the gland, partial removal of the anterior lobe (partial hypophysectomy) was performed (see Operative results: Patients with and without MRI detection of adenoma).

Histological analysis

Resected specimens, including adenoma, a portion(s) of the anterior lobe, and/or dura (when invasion was suspected) were analyzed using hematoxylin and eosin staining, as well as reticulin staining and ACTH immunohistochemistry.

Statistical analysis

Statistical analyses were performed with SPSS Statistics 19 software (IBM Corporation, Somers, New York). Variables were analyzed for impact on initial and lasting remission using binary logistic regression or Fischer's exact test. Measurable variables were analyzed for correlation using linear regression analysis and means compared with Student's independent t test. Multivariate binary logistic regression was performed on those variables that achieved a P value of .1 or less on univariate testing. Categorical variables were analyzed for correlation using χ² test.

Results

Patient characteristics

Two hundred consecutive CD patients (94 males, 106 females) were included (Table 1). Twenty-seven patients (13%) had prior surgery at another institution. Mean age of symptom development was 10.6 ± 3.6 years (range, 4 to 19 y; all but 1 patient had the onset of symptoms at the age of 18 or earlier) (Table 2). Mean preoperative body mass index was 31.0 ± 7.4 kg/m² (range, 16.4 to 61.0 kg/m²), with 140 of 198 patients (71%) classified as obese (11). Mean age at NIH surgery was 13.7 ± 3.7 years (range, 4.5 to 20.8 y); 91 children had not reached puberty, and 109 had. Mean interval from initial symptom to surgery was 3.3 ± 2.1 years (median, 3.0 y; range, 0 to 13 y). Female-to-male ratio was significantly different (χ², P = .000006) between patients who had surgery before the age of 15 years (.8:1) and patients who had surgery after that age (3.7:1) (Figure 1). The only symptoms and signs with a different incidence in children before and after puberty were decreased linear growth, abdominal striae, and easy bruising (Table 2).

Table 1.

Results of Preoperative Biochemical Testing

Test	Positive test, n (%)
Establish Cushing's syndrome
24-h UFCa	177 of 179 patients (99)
24-h UFC/meters2 of body surface areab	155 of 177 patients (88)
Loss of diurnal serum cortisol variationc	188 of 193 patients (97)
Differential diagnosis of Cushing's syndrome
HDDSTd	151 of 167 patients (90)

^aUFC elevated when compared to age-appropriate reference.

^bElevated if > 70 μg/m².

^cLoss of variation defined as PM serum cortisol > 7.5 μg/dL.

^dHDDST, high-dose dexamethasone suppression test, positive when AM cortisol was suppressed by at least 75%.

Table 2.

Presenting Signs and Symptoms

Presenting Signs and Symptoms	Prepubertal Patients (n = 91)	Postpubertal Patients (n = 109)	All Patients (n = 200)
Rapid weight gain	88 (97)	98 (90)	186 (93)
Obesitya	81 (89)	59 (54)	140 (71)
Dorsal cervical or supraclavicular fat pad	58 (64)	79 (72)	137 (69)
Moon facies	56 (62)	70 (64)	126 (63)
Decreased linear growthb	77 (85)	48 (44)	125 (63)
Central obesity	57 (63)	66 (61)	123 (62)
Hirsutism	47 (52)	65 (60)	112 (56)
Abdominal striaeb	39 (43)	70 (64)	109 (55)
Fatigue	43 (47)	52 (48)	95 (48)
Acne	44 (48)	50 (46)	94 (47)
Amenorrhea (primary or secondary)c	NA	49 (72)	52 (49)
Headaches	33 (36)	43 (39)	76 (38)
Hypertension	27 (30)	44 (40)	71 (36)
Acanthosis nigricans	32 (35)	32 (29)	64 (32)
Depression, anxiety, mood swings	22 (24)	39 (36)	61 (31)
Easy bruisingb	14 (15)	35 (32)	49 (25)
Gynecomastiac	8 (15)	9 (22)	17 (16)
Glucose intolerance or diabetes	2 (2)	11 (10)	13 (7)
Alopecia	3 (3)	8 (7)	11 (6)
Bone fractures	3 (3)	4 (4)	7 (4)
Nephrolithiasis	1 (1)	6 (6)	7 (4)

Abbreviation: NA, not applicable. Data are expressed as number of patients (percentage).

^aDefined as a body mass index at or above the 95th percentile for children of the same age and sex (12).

^bSignificantly different (Fisher's exact test, P < .01).

^cGender-specific (percentage of females ≥ 10 years old at initial presentation or at surgery; percent of males).

Figure 1.

Gender ratio vs age at surgery.

Preoperative studies

Results of endocrine testing to establish Cushing's syndrome and the underlying source of cortisol excess are shown in Table 1.

Inferior petrosal sinus sampling

Bilateral simultaneous IPSS, which required general anesthesia in the youngest children, was successfully achieved in all 140 patients attempted (age range, 5.8 to 20.8 y). It was diagnostic for a pituitary source of ACTH production in 139 patients (99%; venography of the sole inaccurate result demonstrated no anatomic or catheter positioning abnormalities). IPSS lateralization of ACTH (side-to-side ratio over 1.4) occurred in 123 patients (88%). It accurately predicted the half of the gland containing the adenoma in 57 of the 82 patients (70%) in whom a discrete adenoma off the midline was identified at surgery.

Magnetic resonance imaging

A total of 193 patients (97%) underwent contrast-enhanced MRI. Eighty patients (41%) were imaged only with standard postcontrast spin echo T1 sequences, and 113 patients (59%) underwent standard sequences and SPGR sequences. MRI was positive for an adenoma in 97 patients (50%). SPGR sequences identified an adenoma that was otherwise not visible by spin echo sequences in 17 of the 113 patients imaged with both sequences (15%) (Figure 2).

Figure 2.

Pituitary MRI obtained after contrast with standard spin echo (left panel) and SPGR (right panel) techniques. Note microadenoma on the right side of the gland at the top of the anterior lobe seen with the SPGR, but not with the spin echo scan.

Mean diameter of magnetic resonance-visible adenomas was 6.9 ± 4.3 mm (range, 2 to 22 mm). Adenoma location at surgery corresponded with MRI in 89 patients (92%). For adenomas identified surgically, mean adenoma size (7.2 ± 3.5 mm; range, 2 to 22 mm) was larger in patients with positive imaging than those without positive imaging (mean, 5.1 ± 2.9 mm; range, 1 to 15 mm) (independent 2-sided t test; P = .0004), but significant size overlap existed (Figure 3). Adenoma size predicted by imaging and size at surgery were similar (linear regression; R² = .77).

Figure 3.

Size of adenomas, as measured at surgery, in patients with positive and negative MRI. Although adenomas detected on MRI were larger than adenomas with negative MRI (independent 2-sided t test; P = .0004), there was considerable overlap in the size of the adenomas in the two groups, which suggests that factors other than size, such as tissue consistency of other unknown factors, account for a negative MRI in many patients. The bars represent the mean diameter.

MRI correctly predicted cavernous sinus invasion by adenoma in 7 of 34 patients (21%) and falsely predicted invasion in 1 patient (1%). MRI did not predict noncavernous sinus dural invasion in any of the 29 patients with invasion evident at surgery.

Surgical and pathological findings

Adenomas

A total of 174 patients (87%) had a discrete adenoma identified at surgery. Mean adenoma size in these cases was 6.3 ± 3.3 mm (range, 1.5 to 22 mm), including 156 microadenomas (10 mm or less in diameter; 91%) and 18 macroadenomas (over 10 mm; 9%). Adenomas were identified in the anterior lobe (161 patients; 93%), within the posterior lobe (4 patients; 2%), between the anterior and posterior lobes (6 patients; 3%), within the pituitary stalk (1 patient; 1%), within the cavernous sinus (1 patient; 1%), and in the parasellar sphenoid sinus mucosa (1 patient; 1%). A total of 158 adenomas (91%) stained for ACTH. Macroadenomas were found in 7.4% of children < 13 years old, all of whom were prepubertal, and 10.1% of children ≥ 13 years old at surgery (not significant; Fisher's exact test). In 26 patients (13%), no adenoma was surgically identified (all had negative MRI), and partial hypophysectomy was performed. An ACTH-positive adenoma was found within resected pituitary tissue in 8 of these patients (31%). No patient had a diagnosis of corticotroph hyperplasia. Although the incidence of negative surgical exploration was higher in children <13 years old (17.2%) vs children ≥ 13 years old (9.2%), the difference was not significant (P = .12; Fisher's exact test).

Dural invasion

Forty-three patients (22% of all patients) had parasellar dural invasion (33 patients without [19%] and 10 with [37%] prior surgery; χ², P = .035). Fourteen patients had only cavernous sinus, 20 had invasion of the cavernous sinus and adjacent sella dura, and 9 had only noncavernous sinus sella dura invasion. There was no difference in parasellar invasion in children < 13 years old (16.0%) vs children ≥13 years old (23.5%; not significant, Fisher's exact test).

Complications

The most common complication was postoperative diabetes insipidus at discharge in 10 patients (5%). Three patients (1.5%) developed seizures (0 to 2 d after surgery) due to hyponatremia, which resolved with sodium normalization. Three patients (1.5%) had maxillary fractures caused by intranasal retractors (2 had transient postoperative diplopia treated conservatively with resolution after 1 d and 2 mo; 1 patient had unilateral ophthalmoplegia and vision loss treated with surgical decompression of the fracture fragment with return of normal vision and extraocular motor function). One patient developed a delayed cerebrospinal fluid leak 2 days after surgery, which was treated with surgery and lumbar cerebrospinal fluid drainage.

Eight patients developed side effects associated with successful remission of CD. Six (2%) developed pseudotumor cerebri in the delayed postoperative period as part of the steroid withdrawal syndrome (12). One patient developed transient somnolence on postoperative day 2 that coincided with angiographic evidence of cerebral vasculitis with thalamic and hypothalamic infarcts. One patient developed a transient posterior encephalopathy syndrome on postoperative day 1 that presented with hemiparesis and scattered white matter lesions on MRI. Deficits in these patients resolved with steroid replacement. We do not have sufficient data to reliably address the status of overall pituitary function of the normal gland after surgery.

Operative results

Early biochemical remission

A total of 191 patients (96%) underwent a single NIH operation, whereas 9 patients (4%) failed to achieve immediate postoperative remission and underwent early reoperation, which was successful in 6 of the patients. Overall, 195 of the 200 patients (98%) achieved remission after surgery (170 of 173 patients [98%] without prior surgery, and 25 of the 27 patients [93%] with prior surgery). One hundred eighty-nine of the patients (97%) that achieved remission were hypocortisolemic, and 6 patients (3%) were eucortisolemic. Five patients (2%) did not achieve remission after surgery, 2 then underwent successful radiation therapy, one received bilateral adrenalectomy, one was treated with medical therapy, and one was lost to follow-up.

Lasting biochemical remission

Follow-up was available in 179 patients (90%; mean postoperative follow-up, 6.8 ± 4.7 y; median, 6.0 y; range, .3 to 21.3 y). Two patients, both with prior surgery, underwent early postoperative adjuvant radiation therapy for extensive unresectable parasellar dural invasion. Overall, 14 patients (8% of patients with at least 1 y follow-up) had recurrence of disease after surgery (mean time to recurrence, 4.7 ± 3.8 y; range, .7 to 11.7 y), including 12 patients (8%) without and 2 patients (8%) with prior surgery (4). Overall, patients with early biochemical remission had recurrence-free survival rates at 1 and 5 years after surgery of 99 and 92%, respectively (Figure 4A). There was no difference in recurrence rates between the groups ≤13 years old (3 of the 91 patients ≤ 13 years old had undergone puberty) and > 13 years old (3 of the 109 patients > 13 y old were prepubertal) (see Figure 4B).

Figure 4.

A, Kaplan-Meier plot of recurrence-free survival in patients with remission of CD after surgery. B, Comparison of recurrence-free survival between patients ≤ 13 years old and patients > 13 years old. Below the abscissa, 80/94 indicates the number of patients at risk ≤ 13 years/>13 years old at each interval.

Patients with and without MRI detection of adenoma

An adenoma was identified on MRI in 97 patients. In 96 of these patients a discrete lesion considered to be adenoma was identified at surgery (99%). MRI accurately defined the site of an adenoma at surgery in 89 of these 97 patients (92%) and 87 of those 89 patients (98%) were hypocortisolemic after surgery. In 8 patients the MRI was interpreted as positive but did not correlate with the surgical findings; in 7 of these patients an ACTH-staining adenoma was discovered elsewhere during exploration of the gland and one patient underwent partial hypophysectomy with an ACTH-staining adenoma in the resected tissue. All 8 patients were hypocortisolemic after surgery. Despite a negative preoperative MRI, in 72 of the 96 patients (75%), the surgeon identified an adenoma, 60 of which (83%) proved to be ACTH-positive adenomas. In 24 of the 96 patients (25%) with negative MRI, an adenoma could not be identified, and a portion of the pituitary, usually about half of the anterior lobe on the side corresponding to an ACTH gradient from IPSS, was removed. Although these specimens contained an adenoma in 7 patients (29%), 22 patients had remission of CD. Overall, 94 of the 96 children with negative pituitary MRI (98%) had remission after surgery.

Factors associated with successful surgery and recurrence

One hundred seventy-one of 174 patients (98%) that had an adenoma identified at surgery achieved early biochemical remission. Two of the 3 patients without early biochemical remission had extensive parasellar invasion that could not be completely removed. One patient who had a negative MRI underwent resection of a 3-mm lesion that appeared consistent with an adenoma but did not stain for ACTH. One hundred fifty-six of 158 patients (99%) with an ACTH-staining adenoma in the tissue from surgery had early biochemical remission. The 2 failures had tumor filling the cavernous sinus. Remission occurred in 24 of 26 patients (92%) who received partial hypophysectomy.

Persistent hypercortisolism (surgical failure) occurred in 3 patients without prior surgery (2%). One patient had extensive tumor within the cavernous sinus that was not completely resectable. The other 2 patients had no tumor identified, and partial hypophysectomy did not result in biochemical remission. In the remaining 24 patients with cavernous sinus invasion, remission was achieved by excising the medial cavernous sinus dural wall, as well as free tumor within the sinus. Similarly, dural invasion at other sites was managed with resection of the region of invaded dura.

Of the 14 patients with recurrent hypercortisolism during follow-up, 9 patients (64%) had 1 or more risk factors for recurrence. This included 5 patients (36%) with cavernous sinus and contiguous dural invasion, 2 patients with a macroadenoma (14%), 2 patients (14%) who had no distinct adenoma and underwent partial hypophysectomy (including one that had an ACTH-staining adenoma), and 2 patients (14%) with presumed adenomas resected that did not stain for ACTH. Thirteen of these 14 patients (93%) had hypocortisolism, and one (7%) was eucortisolemic immediately after surgery.

Univariate predictors associated with lasting remission were identification of adenoma at surgery (irrespective of invasion; P = .03) and noninvasive ACTH-staining adenoma (P = .05). All 110 patients (100%) with a noninvasive ACTH-staining adenoma identified at surgery had remission of CD (108 hypocortisolemic, 2 eucortisolemic), and 94% remained recurrence-free at follow-up.

Preoperative and intraoperative factors were analyzed for association with early biochemical remission and lasting remission (Table 3). Sensitivities, specificities, and predictive values associated with lasting remission were calculated for various endocrinological result thresholds (Table 4). Analysis with receiver operating characteristics was used to assess the value of the nadir morning serum cortisol, 24-hour UFC, and morning serum ACTH to predict recurrence-free survival. However, no significant threshold values of any measure were identified, and the receiver operating characteristic plot area under the curve (AUC) was < .6 for all. Although the highest positive predictive value (PPV) for lasting remission was provided by a morning serum cortisol nadir less than 1.0 μg/dl, 5 of the 14 recurrences occurred among the 115 patients with a minimum morning serum cortisol of < 1.0 μg/dl (Figure 5), and high PPV came at the expense of low specificity (Table 4).

Table 3.

Factors Associated With Early and Lasting Biochemical Remission^a

	P Value for All Patients (n = 200)	P Value for Patients Without Prior Surgery (n = 173)	P Value for Patients With Prior Surgery (n = 27)
Factors associated with early biochemical remission
Lack of prior surgery	.1	N/Ab	N/Ab
Age at symptom onset	.9	1.0	.7
Age at surgery	.9	.6	.9
Pubertal status	.3	.6	.4
Gender	.6	.6	.2
Preoperative morning cortisol	1.0	.9	1.0
Preoperative 24-h UFC	.4	.3	.6
Preoperative high-dose dexamethasone suppression	.7	.9	N/Ac
Visibility on MRI	.6	.6	.9
Adenoma identified at surgery	.1	.03	.8
Adenoma located centrally within gland	.7	.8	.9
Adenoma size	.3	1.0	.5
Lack of cavernous sinus invasion	.2	.4	.6
Lack of other parasellar dural invasion	.7	.3	.5
ACTH+ adenoma	.006	.05	.04
Noninvasive ACTH+ adenoma	.008	.05	.3
Factors associated with lasting biochemical remission
Lack of prior surgery	1.0	N/Ab	N/Ab
Younger age at onset of symptoms	.03/.03d	.05	.4
Age at surgery	.4	.6	.4
Pubertal status	.1	.3	.4
Gender	.9	.7	.6
Preoperative morning cortisol	.4	.3	.8
Preoperative 24-h UFC	.7	.6	.3
Preoperative high-dose dexamethasone suppression	.8	.8	N/Ac
Negative MRI	.09	.2	.3
Adenoma identified at surgery	.9	.7	.8
Adenoma located centrally within gland	.1	.1	.9
Smaller adenoma size	.005/.4d	.04	.09
Lack of cavernous sinus invasion	.04/.6d	.2	.08
Lack of other parasellar dural invasion	.02/.3d	.2	.06
ACTH+ adenoma	.7	.5	.7
Noninvasive ACTH+ adenoma	.1	.3	.2

Features with statistical significance (P < .05) are indicated in bold.

^aAnalysis performed with binary logistic regression, or Fischer's exact test for groups with less than 1 event.

^bNot applicable (N/A), given all patients in group either had or did not have prior surgery.

^cNot applicable (N/A), given all patients demonstrated suppression on testing.

^dMultivariate analysis.

Table 4.

Accuracy of Early Postoperative Endocrine Tests in Predicting Lasting Remission

Test Threshold	Sensitivity (%)	Specificity (%)	PPV (%)	NPV (%)
Nadir AM cortisol <1.0 μg/dL	64	69	96	15
Nadir AM cortisol <2.0 μg/dL	93	23	93	23
Nadir AM cortisol <3.0 μg/dL	97	15	93	33
Nadir AM cortisol <5.0 μg/dL	98	8	92	25
Nadir AM ACTH <5.0 pg/mL	70	50	95	12
Nadir AM ACTH <10.0 pg/mL	96	25	94	33
Nadir AM ACTH <15.0 pg/mL	98	25	94	50

Abbreviations: AM, morning; NPV, negative predictive value.

Figure 5.

Correlation of measures of cortisol production on days 3–5 after surgery and recurrence or sustained remission during follow-up is shown for various categories of patients. The minimum values of AM cortisol (A) and 24-hour UFC (B) from days 3–5 after surgery are shown for each patient on the ordinate, and the duration of follow-up for that patient after surgery is presented on the abscissa. The 5 patients who failed surgery and patients with no follow-up are not included. Red markers indicate recurrence. A, Minimum AM cortisol on days 3–5 after surgery. B, Minimum 24-hour UFC on days 3–5 after surgery presented as a fraction of the lower limit of the normal range at the assay contemporary with the surgery (the range of normal changed with the use of different assays during the study; see “Patients and Methods”) and the patient age. The numbers in parentheses in panel B are the minimum AM cortisol for that patient. The limited number of patients with values exceeding very low levels in panels A and B limits the capacity to compare the use of those values to predict recurrence. Note also the limited duration of follow-up in most of the patients with measures of cortisol production in the normal range (eucortisolism). Measurements below the detection limit of the assay are assigned a value equal to the lower limit in panels A and B.

Discussion

The adverse effects of hypercortisolism during development and the potential for lifelong effects of pediatric CD impart importance to its successful diagnosis and treatment. Because selective resection of ACTH-secreting adenomas produces immediate remission while maintaining pituitary function, surgery is the initial treatment. Several previous manuscripts describe pediatric CD and its treatment (1, 2, 13–16). However, the small number of patients reported (all include ≤ 50 patients) limits the finality of their observations. We defined the clinical characteristics, examined factors influencing the outcome of surgical management, and assessed the accuracy of postoperative testing to predict lasting remission in a large series of children with CD.

Although adults frequently present with obesity, hypertension, and diabetes, common features of pediatric CD include rapid weight gain associated with decreased linear growth (Table 1) (1, 2, 17). Early diagnosis and successful treatment avoid the stunted linear growth associated with prolonged childhood CD. Since the first publications describing treatment of pediatric CD, it has been evident that the ratio of boys and girls is about equal when the disease onset is prepubertal (2, 17). Why adult CD affects females preferentially (18, 19) is unknown. It is probably associated with adult female hormonal status because in this series, which included a large number of pediatric patients affected before and after puberty, thus providing assessment of that transition, the female to male ratio in CD children less than 15 years old was .8:1, whereas in children after age 15 the ratio was 3.7:1, similar to the known adult gender distribution.

MRI with SPGR sequences detected adenomas in 15% of patients that were not detected by standard spin echo sequences. More sensitive imaging of the adenoma in CD gives several advantages because positive MRI helps confirm the diagnosis of CD, potentially avoids the risks of IPSS, and permits focused pituitary exploration at the MRI location of the adenoma. For instance, a discrete adenoma was found at surgery in 96 of the 97 patients with a positive MRI (99%), and 95 of those 96 patients (99%) were hypocortisolemic (90 patients) or eucortisolemic (5 patients) after surgery. On the other hand, even with negative preoperative MRI, the pituitary adenoma can be found at surgery in most patients (75% of the 96 patients with negative MRI), and most patients with negative MRI have remission in response to selective adenomectomy or partial hypophysectomy (94 of the 96 patients, or 98%). These findings are in contrast to the observations of the limited accuracy of MRI for predicting dural invasion (20) (cavernous sinus invasion was detected by MRI in only 21% of the 34 affected patients), which was detected in 22% of patients without previous surgery and which involved the cavernous sinus in 79% of patients with dural invasion.

IPSS is the most accurate diagnostic test for CD in Cushing's syndrome (6). All 140 children in whom it was attempted had successful sampling from both IPSSs without adverse events. Almost all diagnostic failures with IPSS have been false-negative findings in patients with CD (5). The results of IPSS were consistent with a diagnosis of CD in 139 of the 140 children (99%) who underwent the test, although these results cannot be used to evaluate the diagnostic accuracy of IPSS because children with other etiologies of Cushing's syndrome were not evaluated. Additionally, the use of IPSS side-to-side ratio to determine adenoma lateralization within the pituitary gland was accurate in only 72% of patients without prior surgery, which is comparable to the findings in a large series of adult patients studied at the same center (6) and 55% of patients with prior surgery. Thus, MRI was substantially more accurate than IPSS for determining tumor location.

Ninety-eight percent of patients without and 93% of patients with prior surgery had early hypocortisolism after initial NIH surgery or early repeat surgery, and early hypocortisolism predicts long-term biochemical remission. Factors associated with early remission of CD included identification of a discrete adenoma at surgery, resection of an ACTH-staining adenoma, and absence of dural invasion (Table 3).

Repeat surgery for pediatric CD is an effective management strategy. After previous experience with surgical management of patients with prior surgical failure or recurrent CD (21, 22), we concluded that certain selection criteria can be used to enhance the likelihood of a successful outcome of repeat surgery in this setting. Those criteria include factors that permit knowledge of where the adenoma is likely to be (a positive MRI, knowledge of the location of the adenoma based on the prior surgical report, and/or positive histology for an ACTH-staining adenoma). Twenty-seven patients in the current series had undergone prior unsuccessful surgery at an outside institution, and early biochemical remission was achieved in 93%. Remission was lasting in 92% of those with initial postoperative UFC less than 20 μg/d or morning cortisol of 5 μg/dL or less (Figure 5). The decreased rate of remission in this patient population probably can be attributed to the increased rate of cavernous sinus and noncavernous sinus dural invasion in this group.

The results in the current study compare favorably with those of other reports of the outcome of surgery for pediatric CD (summarized in Ref. 15). There are advantages of using the histological pseudocapsule of an adenoma as a surgical capsule in the excision of pituitary adenomas (10). The results in patients with CD have been described (23). Furthermore, in patients with relatively rare disorders, such as pediatric CD, enhanced outcomes derive from evaluation and treatment at centers with substantial experience (16). Overall, 92% of children with initial remission and follow-up of at least 1 year had sustained remission at last follow-up. To determine early predictors of long-term biochemical remission, we analyzed potential early biological markers of long-term control. Specifically, we calculated the PPV of early postoperative laboratory values related to cortisol regulation, including serum cortisol nadir, UFC nadir, ACTH nadir, and CRH stimulation cortisol levels. All these tests had highly predictive associations (greater than 90%) with lasting biochemical remission (Table 4).

Because lasting remission is the objective of surgery, it is important to establish the factors that are associated with it (Table 3). The association of younger age, smaller tumor size, and absence of dural invasion with lasting biochemical remission suggests that earlier diagnosis when the tumor is small and noninvasive will enhance long-term outcome. Finally, our results, in general, demonstrate that pituitary surgery by an experienced neurosurgeon is the initial treatment of choice in children with CD.