Abstract

Materials and Methods

This retrospective cohort study was approved by the Mayo Clinic Institutional Review Board and included adult patients diagnosed with CS with concurrent hyperglycemia treated with a curative procedure between January 1, 2000, and November 1, 2019. Patient consent waiver was used for this project excluding those who did not give research authorization. Patients were included if they were diagnosed with endogenous CS based on clinical evaluation by the endocrinologist and diagnosis of CS according to the most recent guidelines [1]. Hyperglycemia was classified as DM and prediabetes (impaired glucose tolerance or impaired fasting glucose) diagnosed based on the American Diabetes Association guidelines [23]. All included patients had undergone a curative procedure for treatment of CS (transsphenoidal resection of pituitary adenoma, unilateral or bilateral adrenalectomy, resection of ectopic ACTH-secreting tumor, radiation therapy) and demonstrated remission from CS (defined as low cortisol postoperatively in all patients except in those undergoing bilateral adrenalectomy). Patients needed to have at least one available assessment for hyperglycemia during CS remission. Because patients develop adrenal insufficiency and glucocorticoid withdrawal syndrome following the curative procedure for CS that may last months or be permanent, we have not prespecified a specific required glucocorticoid dose at the end of follow-up, but performed a subgroup analysis based on hydrocortisone dose. We have used the 2 arbitrary cutoffs of 10 and 12 mg of hydrocortisone/m² body surface area for subgroup analysis. Eligible patients were identified through review of medical records of patients with the diagnosis of CS (Fig. 1).

Open in a separate window

Figure 1.

Flowchart for inclusion of patients in the study. The initial search revealed 1479 patients at Mayo with the possible diagnosis of Cushing syndrome (CS). The medical record was reviewed to confirm the diagnosis of endogenous CS and hyperglycemia. The final cohort was determined based on patients who met our inclusion/exclusion criteria.

Medical records were reviewed and data on patient demographics, signs, and symptoms of CS, laboratory data, imaging data, the subtype of CS (pituitary, ectopic, adrenal) were collected. Based on the available clinical and laboratory data, each patient was assigned a biochemical and clinical severity of CS classification score, as previously described [24] (Supplementary Tables 1 and 2 [25]).

Data on hyperglycemia included biochemical data (fasting glucose, glycated hemoglobin A_1c [HbA_1c]) and medications used for management of DM. Data were collected before the curative procedure (baseline), and at any available postoperative time points for each individual patient. We collected hyperglycemia data during glucocorticoid taper, with the final postoperative workup when the patient was off glucocorticoids completely, or on stable glucocorticoid chronic replacement therapy after bilateral adrenalectomy.

Our primary end point was improvement or resolution of hyperglycemia following remission of CS with subgroup analyses based on the CS subtype. Resolution of hyperglycemia was defined as the absence of hyperglycemia (based on HbA_1c and/or fasting glucose, as well as review of the patient’s glucose log at the end of follow-up), without any need for antihyperglycemic therapy. Improvement was defined as one of the following: 1) improvement in HbA_1c and/or fasting glucose concentrations at the end of follow-up while on stable antihyperglycemic therapy; or 2) the decision to decrease the dose and/or number of antihyperglycemic medications/insulin by the treating physician from whom quantitative data were not available (eg, provider notes commenting on glucometer measurements and/or other clinical parameters used to make this judgment).

Secondary end points included changes in HbA_1c (Δ HbA_1c) and the intensity of hyperglycemia management (number of oral medications, total dose of insulin) during follow-up compared to baseline. Subgroup analysis for Δ HbA_1c included 1) patients on stable management for hyperglycemia, 2) patients who needed a decrease in the intensity of hyperglycemia management, and 3) patients who needed an increase in the intensity of hyperglycemia management during the follow-up that followed the curative procedure.

Results

Hyperglycemia After Curative Procedure for Cushing Syndrome

Patients were followed for hyperglycemia management for a median of 10.5 months (range, 0-130 months) after achieving CS remission. At the time of final follow-up, 102 patients (58.6%) were either no longer treated with glucocorticoids or reached a stable glucocorticoid replacement dose. Median hydrocortisone mg/m² body surface area was 8.1 (interquartile range = 0-13.4) at the time of final follow-up.

Subtype of hyperglycemia, biochemical severity of hypercortisolism, and subtype of hypercortisolism were associated with improvement of hyperglycemia during follow up (Fig. 2A-2C). When compared to baseline, at the time of final follow-up, mean HbA_1c had decreased by a mean of 0.84%; P ≤ .001 (Fig. 2D). The degree of HbA_1c decrease (Δ HbA_1c) was associated with weight loss (Δ body mass index [BMI]) (Fig. 2E). In 60 patients treated with one or more noninsulin agents at baseline, the overall number of medications decreased at follow-up by a mean of 0.5 agents (from mean of 1.3 to mean of 0.76, P ≤ .001) (Fig. 2F). In 42 patients treated with insulin at baseline, the mean total daily insulin use decreased by a mean of 30 units (P ≤ .001) Fig. 2G). In 33 patients treated with insulin at baseline and follow-up, the mean total daily insulin/kg dose decreased by –0.19 units/kg/day (95% CI, –0.02 to –0.36) (P = .02).

Open in a separate window

Figure 2.

Baseline and postoperative glycated hemoglobin A_1c (HbA_1c), number of oral hyperglycemic medications, and total daily insulin units. A, Improvement of hyperglycemia based on the subtype of hyperglycemia. B, Improvement of hyperglycemia based on the biochemical severity of hypercortisolism. C, Improvement of hyperglycemia based on the subtype of hypercortisolism. D, HbA_1c at baseline and at follow up (n = 130). E, Association of the Δ HbA_1c with Δ body mass index following the curative procedure for Cushing syndrome. F, Number of oral medications at baseline and at follow-up in 93 patients initially treated with oral medications for hyperglycemia. G, Total daily insulin units at baseline and at follow up (n = 64).

At the time of final follow-up, hyperglycemia improved in 118 patients (68.6%) (Table 4). Overall, HbA_1c decreased gradually during follow-up (Fig. 3A). In patients in whom no changes in therapy for hyperglycemia during follow-up was instituted (n = 42), HbA_1c decreased by a mean of –0.5% (median of –0.3%) by the end of follow-up (Fig. 3B). In 71 patients in whom the intensity of hyperglycemia management had to be decreased during follow-up, HbA_1c decreased by a mean of –1.2% (median of –1.0%) (Fig. 3C). Only 17 patents had to intensify hyperglycemia management after curative surgery and demonstrated a mean Δ HbA_1c of +0.2% (median of +0.1%) (Fig. 3D). This cohort consisted of 9 patients with pituitary CS (52.9%), 7 patients with adrenal CS (41.2%), and 1 patient with ectopic CS (5.9%). Of these 17 patients, 5 (29.4%) had severe CS, and baseline median HbA_1c was 6.9% (5.5%-10.7%). Univariate and multivariable analysis did not demonstrate a significant factor in this group that was associated with the need to increase therapy (data not shown). Notably, no differences in the initial postoperative daily glucocorticoid dose, in the number of patients still treated with glucocorticoids, or in the median daily hydrocortisone dose at the end of follow-up were found between the groups based on the intensity of hyperglycemia management (Supplementary Table 6 [25]).

Open in a separate window

Figure 3.

Δ Glycated hemoglobin A_1c (HbA_1c) following the curative procedure for Cushing syndrome. A, Δ HbA_1c in all patients (n = 130). B, Δ HbA_1c in 42 patients in whom hyperglycemia management was not changed (including those on no medications). C, Δ HbA_1c in 71 patients in whom the intensity of hyperglycemia management was decreased (decrease in dose or number of medications). D, Δ HbA_1c in 17 patients in whom the intensity of hyperglycemia management was increased (increase in dose or number of medications, new start of medications).

In a subgroup analysis of patients who discontinued glucocorticoid therapy at the last follow-up compared to patients who were still being treated with glucocorticoids, or based on the final daily hydrocortisone cutoff of 10 and 12 mg/m² body surface area, no differences in the decrease of HbA_1c or the proportion of patients with improved hyperglycemia were observed (Supplementary Table 7 and 8 [25]). In another subgroup analysis of patients treated with bilateral adrenalectomy compared to patients treated with other procedures, no differences in the decrease of HbA_1c or proportion of patients with improvement of hyperglycemia were found (Supplementary Table 9 [25]).

On univariate analysis, age, sex, and BMI were not associated with improvement of hyperglycemia following CS remission (Table 4). In contrast, higher severity of CS based on the biochemical and clinical severity scores, ACTH-dependent vs -independent CS, and DM2 vs prediabetes were associated with improvement of hyperglycemia (see Table 4 and Fig. 2A-2C). Severity of CS (OR = 2.4; 95% CI, 1.1-4.9) and ACTH-dependent CS (OR = 2.9; 95% CI, 1.3-6.4), but not the type of hyperglycemia (DM vs prediabetes: OR = 2.13 [95% CI of 0.9-4.9]), were found to be associated with hyperglycemia improvement on multivariable analysis (see Table 4).

Table 4.

Factors associated with improvement in hyperglycemia after curative procedure^a

Univariate analysis				Multivariable analysis
Variable	Improvement in hyperglycemia N = 118	No improvement in hyperglycemia N = 54	P	Variable	Odds ratio (95% CI)	P
Age, y, median (range)	52.5 (19-82)	54 (28-78)	.34
Female sex, n (%)	85 (72%)	41 (75.6%)	.59
BMI, median (range)	34.8 (21.0-66.4)	37 (22.0-53.8)	.76
Duration of hypercortisolism, mo, median (range)	18 (0-240)	12 (0-204)	.49
Duration of hyperglycemia, mo, median (range)	15.5 (0-350)	25 (0-458)	.36
Biochemical severity score, n (%)
Mild	12 (10.1%)	12 (22.6%)	.001	Biochemical severity score b (severe vs other)	2.4 (1.1-4.9)	.02
Moderate	33 (28%)	24 (45.3%)
Severe	73 (61.9%)	17 (30.1%)
Available for n = 173
Clinical severity score, n (%)
Mild	6 (5.1%)	9 (16.7%)	.001
Moderate	31 (26.3%)	24 (44.4%)
Severe	81 (68.6%)	21 (38.9%)
Type of CS, n (%)
Adrenal	19 (16.1%)	24 (44.4%)	< .001	Type of CS (ACTH-dependent vs adrenal)c	2.9 (1.3-6.4)	.007
Ectopic	18 (15.3%)	6 (11.2%)
Pituitary	81 (68.6%)	24 (44.4%)
Hyperglycemia diagnosis, n (%)
Prediabetes	16 (13.6%)	17 (31.5%)	.006	Hyperglycemia diagnosis (diabetes vs prediabetes)	2.13 (0.9-4.9)	.08
DM2	102 (86.4%)	37 (68.5%)

Open in a separate window

Abbreviations: ACTH: adrenocorticotropin; BMI, body mass index; CS, Cushing syndrome; DM, diabetes mellitus.

^a Patients with DM1 were excluded (n = 2).

^b Because biochemical and clinical severity scores were concordant, only one was chosen for the multivariable analysis.

^c Pituitary and ectopic CS were combined into ACTH-dependent subtype for the multivariable analysis.

Discussion

In this study, we characterized the presentation at baseline and the evolution of hyperglycemia in patients with CS following the curative procedure and CS remission. We showed that two-thirds of patients demonstrated resolution or improvement of hyperglycemia during follow-up, and more than half were able to decrease the dose or the number of medications used for hyperglycemia. Patients with ACTH-dependent CS and those with a higher biochemical severity score were more likely to have improved hyperglycemia during follow-up.

We found that at the time of diagnosis, no differences in baseline HbA_1c or fasting glucose were found between the CS subtypes; however, patients with pituitary CS were more commonly treated for hyperglycemia and with a higher number of hyperglycemia agents than patients with other CS subtypes. Potential explanations for this finding include a higher duration of hypercortisolism in patients with pituitary CS before diagnosis, along with a high proportion of patients with severe biochemical and clinical CS severity scores. Patients with pituitary CS also had the highest BMI, further contributing to the severity of hyperglycemia at baseline. While several studies reported on the differences in the prevalence of hyperglycemia among CS subtypes [2, 4, 6, 15], none commented on the severity and the intensity of therapy of hyperglycemia at the time of diagnosis based on CS subtype.

We found that almost half of patients in our study demonstrated improvement in hyperglycemia, and another 21% experienced resolution of hyperglycemia following the curative procedure for CS. Several small studies in patients with CS reported resolution of hyperglycemia in 20% to 79% of patients [14-17, 19-22]. Improvement in hyperglycemia following adrenalectomy was reported in 52% of patients with mild autonomous cortisol secretion [26], but was not systematically reported for patients with CS. In our study we defined improvement as a decrease in HbA_1c and/or a decrease in the dose or number of medications used to treat hyperglycemia. Applying this definition, 47% of our patients demonstrated improvement in hyperglycemia by the end of follow-up. We showed that the overall HbA_1c decreased by 0.84%, and only a minority of patients needed to intensify their hyperglycemia regimen after achieving CS remission.

We identified several factors that were associated with either resolution or improvement of hyperglycemia. These included the severity of CS (both clinical and biochemical), type of CS, and severity of hyperglycemia at the time of diagnosis. ACTH-dependent CS and severe biochemical score were independent predictors of hyperglycemia improvement on multivariable analysis. These data suggest that in patients with milder forms of CS, hypercortisolism may play a smaller role in hyperglycemia at the time of diagnosis, and a subsequent curative procedure is less likely to improve it.

Several clinical implications arise from our study. First, as improvement in hyperglycemia is seen in most patients, counseling and close monitoring of those treated with insulin or other medications is needed to modify therapy and avoid hypoglycemia. Second, patients with milder CS (more commonly adrenal) should be treated for more severe hyperglycemia rather than adopt an expectant attitude because hyperglycemia in these cases is less likely to improve with CS remission. Notably, not all patients achieved an optimal glucose control during follow-up, suggesting that intensification of therapy rather than an expectant attitude is warranted in some patients with CS.

Our study has several strengths. We include a large sample size, with appropriate patient representation regarding CS subtype and severity, and type of hyperglycemia. Availability of a detailed medical record system allowed us to perform an in-depth analysis both of baseline and follow-up variables. This is the first study to analyze the quantitative changes, based on the time from the curative surgery, to assess the changes in the intensity of hyperglycemia therapy and identify predictors for hyperglycemia improvement. The limitations of our study include the retrospective design with selection, referral, and information biases. Not all patients with CS treated at our institution were followed for hyperglycemia after surgery for the whole duration of follow-up. Management of hyperglycemia was not standardized and differed among clinicians. While the initial diagnosis of hyperglycemia was made based on the American Diabetes Association guidelines, resolution of hyperglycemia was based on multiple factors, such as glucose log, normal fasting glucose, and HbA_1c measurements in a patient not treated with pharmacotherapy for hyperglycemia. Duration of follow-up varied, and at the time of last follow-up not all patients recovered adrenal function. Some of the subgroup analyses were limited by the small sample size.

In conclusion, we demonstrated that almost 70% of patients with CS demonstrate either resolution or improvement in hyperglycemia following CS remission. As a group, patients demonstrate a decrease in HbA_1c, and can be treated with less insulin and fewer noninsulin agents. Patients with more severe hyperglycemia, ACTH-dependent CS, and more severe CS are more likely to improve after surgery. Patients with hyperglycemia need appropriate monitoring and medication adjustment after surgery for CS.

Acknowledgments

Glossary

References