Piecing Together the Adrenal Puzzle: CAI Score and Postoperative Testing

Four Studies that Can Improve Patient Outcomes Since April is Adrenal Disease Awareness Month, Endocrine News is highlighting recent studies that shed light on important advances and insights in the field. These studies, all published in Endocrine Society journals, show how improved diagnostic accuracy, postoperative management, and a better understanding of rare adrenal pathologies offer clinicians tools that could improve patient outcomes. At a Glance • The CAI score improves diagnostic accuracy in ambiguous cases of suspected central adrenal insufficiency (CAI) by integrating morning serum cortisol with key clinical parameters and can help guide treatment decisions. • Two-pronged testing consisting of basal cortisol and cosyntropin levels initiated as early as four weeks post-adrenalectomy can identify patients with adrenal insufficiency, guide glucocorticoid treatment, and obviate unnecessary glucocorticoids. • Although typically benign and nonsecretory, adrenal ganglioneuromas (AGNs) can rarely exhibit secretory properties that mimic other adrenal pathologies, making definitive histopathologic assessment essential; in cases of ACTH-dependent hypercortisolism with an adrenal lesion, secretory AGN should remain on the differential. • MEN1-associated adrenocortical carcinoma requires comprehensive hormonal evaluation, regular follow-up for adrenal lesions, and routine screening of asymptomatic mutation carriers for early detection and improved outcomes. Given the adrenal glands’ enormous hormonal reach — governing inflammation, blood pressure, stress response, and more — when they are dysfunctional, the consequences are serious. According to the Endocrine Society’s Endocrine Facts and Figures , the prevalence of primary adrenal insufficiency is estimated at 40 to 100 cases per million in the United States, while Cushing syndrome affects an estimated 8 people per million in those under age 65 years. Rarer still, Cushing disease occurs in roughly 2.3 to 2.7 cases per million per year, and multiple endocrine neoplasia type 1 (MEN1) affects an estimated 3 to 10 people per 100,000. Those numbers may look modest, but behind each one is a clinical journey often fraught with challenges. Two research studies advance our understanding of cortisol testing and postoperative management, while two case reports serve as vivid reminders that the adrenal gland can still confound even the most experienced clinicians. Central Adrenal Insufficiency In “ CAI Score for the Diagnosis of Central Adrenal Insufficiency ,” published in the Journal of the Endocrine Society in February, Mussa H. Almalki, MBBS, MHSc, of the College of Medicine of Alfaisal University in Riyadh, Saudi Arabia, and team have truly moved the needle forward when it comes to day-to-day, in-clinic management. As to what prompted this study, Almalki credits the well-known frustration that comes from wanting to help patients but being hemmed in by existing clinical parameters. “We often see patients who we suspect might have central adrenal insufficiency (CAI) — perhaps they have a pituitary tumor, have had head trauma, or have other hormone deficiencies,” he recounts. “We order a morning cortisol test, hoping for a clear answer. But so often, the result comes back in what we call the ‘gray zone’—typically between 4 and 18 µg/dL. It’s not low enough to confidently diagnose CAI, but it’s not high enough to rule it out.” This dilemma generally creates the need to order a dynamic test (e.g., short Synacthen test [SST]), but this, says Almalki, is time-consuming, requires patients to come to a dedicated unit, can be unpleasant, and is not a perfect test itself. “We realized we needed a better way to stratify risk in these ‘gray zone’ patients before deciding on next steps. We wanted to see if we could combine the information we already had — like the specific cortisol level, the presence of other pituitary issues, and imaging findings — to build a more sophisticated tool than just looking at the cortisol number in isolation. The goal was to help clinicians make a more informed, data-driven decision about who truly needs that dynamic test.” For their retrospective single-center study from a Riyadh tertiary referral center, the team enrolled 341 adults with suspected CAI and indeterminate morning cortisol levels, using the SST as the reference standard for diagnosis. They developed and validated a predictive scoring system that integrates morning cortisol levels, pituitary hormone deficits, tumor size, sex, and treatment history to help stratify CAI risk in the diagnostically challenging “gray zone” where cortisol results alone are inconclusive. Where traditionally, the “gray zone” is defined as about 3 to 15 µg/dL, the team deliberately expanded it to 4 to 18 µg/dL to be useful in the real world, where different labs use different cutoffs. “But more importantly,” explains Almalki, “we know a cortisol of 5 µg/dL isn’t the same as 15 µg/dL, even if both are ‘gray.’ By widening the range, we could let the data show us how risk changes as cortisol drops, rather than forcing it into arbitrary boxes.” The resulting tool — the CAI score — which also incorporates a machine learning model and is freely available as a web-based application, demonstrated stronger diagnostic accuracy than morning cortisol alone. Along the way, the researchers encountered a couple of surprises. The first was what Almalki calls “the sheer power of pituitary hormone deficits.” In their model, having three or more additional hormone deficiencies was a very strong predictor (odds ratio >35). “This really drove home the point that CAI is very rarely an isolated event. It’s often a sign of more widespread pituitary damage. The health of the pituitary gland as a whole is a massive clue to corticotroph function,” he says. The second was the comparatively modest role tumor size played. “While larger tumors did increase the risk, size wasn’t as powerful a predictor as the number of other hormone deficits. This suggests that it’s not just the size of the tumor, but how it’s impacting the function of the surrounding healthy pituitary tissue — as evidenced by the other hormone losses — that truly matters for CAI risk,” explains Almalki. “Endocrinology is a field defined by complex, interacting feedback loops. A single lab value rarely tells the whole story. AI-assisted tools are perfectly suited to integrate multiple data points — labs, imaging, symptoms, other diagnoses — and recognize patterns that are too subtle or complex for the human brain to consistently process. I see these tools not as replacing the endocrinologist, but as powerful allies that will handle the ‘noise’ and allow us to focus on the ‘signal.’” — Mussa H. Almalki, MBBS, MHSc, College of Medicine, Alfaisal University, Riyadh, Saudi Arabia The future for the CAI score certainly looks promising, but two things need to happen for its widespread adoption, according to Almalki: “First, external validation — seeing the model perform consistently across different hospitals and patient populations. Second, demonstrating practical value. If we can show that using the score reduces unnecessary testing, saves money, and doesn’t compromise patient safety, that creates a compelling case for integration into electronic health records or clinical workflows.” The team is not actively pursuing additional related studies, because, as Almalki puts it, “My immediate focus is shifting toward clinical application and direct patient care, rather than driving the next prospective study. I’m happy to leave the door open for other researchers to pick this up and validate it if they see value in it.” As for the importance of integrating AI-assisted tools in medicine, these researchers emphasize such tools are intended to support rather than replace clinical judgment. “Endocrinology is a field defined by complex, interacting feedback loops. A single lab value rarely tells the whole story. AI-assisted tools are perfectly suited to integrate multiple data points — labs, imaging, symptoms, other diagnoses — and recognize patterns that are too subtle or complex for the human brain to consistently process. I see these tools not as replacing the endocrinologist, but as powerful allies that will handle the ‘noise’ and allow us to focus on the ‘signal,’” Almalki says. Finally, for those clinicians similarly frustrated by how to manage a condition in the face of ambiguous results, Almalki has actionable advice: “Stop looking at that gray-zone cortisol in isolation. It’s just one piece of the puzzle. Ask yourself: How low is it? Do they have other hormone issues? What does their MRI show? The CAI score just helps you put those pieces together quickly. A low score might save a patient an unnecessary test. A high score tells you to stop messing around and act. It’s free, it’s fast, and you can use it right now on the website.” [Go to https://cai-predictor.streamlit.app/ .] Post-Adrenalectomy Adrenal Insufficiency In “ Cortisol Testing to Diagnose Adrenal Insufficiency Following Adrenalectomy for Mild Autonomous Cortisol Secretion ,” published just last month in The Journal of Clinical Endocrinology & Metabolism , a team of researchers led by Oksana Hamidi, DO, MSCS, associate professor of medicine of the University of Texas Southwestern Medical Center in Dallas, Texas, and corresponding author Irina Bancos, MD, MSc, professor of medicine and Adrenal Lab Principal Investigator of the Mayo Clinic in Rochester, Minn., sought contemporary data using standardized cortisol thresholds and modern assays on the true prevalence and duration of adrenal insufficiency after adrenalectomy for mild autonomous cortisol secretion (MACS). “Prior studies were heterogeneous, and there remains uncertainty about the optimal postoperative testing strategy,” says Bancos. “Specifically, we aimed to clarify three issues: (1) how often adrenal insufficiency occurs after unilateral adrenalectomy for MACS; (2) whether basal cortisol alone is sufficient, or whether cosyntropin stimulation testing adds value; and (3) whether biochemical/clinical severity correlates with postoperative adrenal suppression and recovery.” The impetus for quantifying these aspects of post-adrenalectomy adrenal insufficiency is patient driven. When recognized and treated appropriately, explains Bancos, this condition is manageable. It becomes problematic, however, when it is unrecognized or unnecessarily prolonged or when patients are empirically treated without appropriate testing. In these situations, patients can experience glucocorticoid withdrawal symptoms, like fatigue, myalgias, and mood changes, that can significantly affect daily functioning. “During this time, patients require education, structured tapering, and close follow-up,” continues Bancos. “Importantly, glucocorticoid withdrawal occurs not due to adrenal insufficiency, but due to abrupt decline in supraphysiologic cortisol before adrenalectomy and lower, more physiologic cortisol levels after adrenalectomy. In this paper, we have not investigated glucocorticoid withdrawal, but we did previously report it in other studies.” Zeroing in on empiric glucocorticoid therapy, why wouldn’t all post-adrenalectomy patients benefit? This team advocates for reserving such treatment for those who truly need it rather than exposing all to steroid hormone exposure with its potential adverse effects — including osteoporosis and bone fractures, increased risk of infections, hyperglycemia/diabetes, weight gain, Cushingoid features, hypertension, myopathy, cataracts/glaucoma, and mental health issues. Their multicenter retrospective study examined 281 patients with MACS, 80% female with a median age of 57 across five U.S. institutions and compared the diagnostic utility of two postoperative cortisol tests: basal cortisol and the cosyntropin stimulation test (CST). Slightly more than half of patients developed postoperative adrenal insufficiency, with younger age ( To account for the increased severity among younger patients, Bancos elaborates: “A possible explanation is that younger individuals may have more biochemically active or more prolonged unilateral disease, resulting in deeper HPA suppression. This requires further study.” Higher BSS also correlated with duration of recovery: Patients with mild scores recovered in approximately three months, whereas those with more severe scores required longer (4–14 months). Age being a strong predictor and the discordancy between tests surprised the researchers, but both findings led to critical insights. “First, basal cortisol and CST were discordant in 22% of patients. This suggests that relying on a single test may misclassify a meaningful subset of patients,” says Bancos. “Second, the strong inverse association with age was unexpected. That challenges prior assumptions and highlights the need for further prospective validation.” Ultimately, their findings could influence endocrinology practice. “In my view,” says Bancos, “adrenal insufficiency is not intolerable when it is anticipated, explained, and carefully managed.” The implication is clear — implement earlier testing. Although the exact optimal timing of testing in the first weeks after surgery remains an area for further research, Bancos says, “our findings support structured reassessment beginning around four to six weeks postoperatively and continuing at regular intervals. Earlier reassessment may prevent unnecessary prolonged glucocorticoid exposure in patients who recover quickly.” Importantly, the scores that come from this earlier testing can inform individualized approaches to management. “Instead of speaking in generalities, we can now say: ‘Based on your biochemical severity, you have a higher likelihood of needing temporary steroid replacement, and recovery may take several months. That improves preparedness and shared decision-making,’” says Bancos. “Prior studies were heterogeneous, and there remains uncertainty about the optimal postoperative testing strategy. Specifically, we aimed to clarify three issues: (1) how often adrenal insufficiency occurs after unilateral adrenalectomy for MACS; (2) whether basal cortisol alone is sufficient, or whether cosyntropin stimulation testing adds value; and (3) whether biochemical/clinical severity correlates with postoperative adrenal suppression and recovery.” — Irina Bancos, MD, MSc, professor of medicine and Adrenal Lab Principal Investigator of the Mayo Clinic, Rochester, Minn. Postoperative stratification is thus clearly beneficial, but what about a validated pre operative risk score? Bancos explains that, while possible, a preoperative score would have built-in limitations. “Biochemical severity and age were strong predictors, yet one critical variable we cannot accurately measure is the duration of cortisol autonomy. MACS is often missed and detected incidentally, making disease duration uncertain,” she says. “While a predictive model could help stratify risk, I do not anticipate that any preoperative score would eliminate the need for postoperative cortisol testing. Objective biochemical assessment remains essential.” The team has several avenues to pursue from here, thanks to their ongoing prospective cohort, which allows them to validate predictors of duration and recovery more rigorously. “Future directions include refining recovery trajectories using longitudinal ACTH and cortisol modeling, exploring mechanistic explanations for age-related differences, developing and validating risk prediction tools, and standardizing postoperative testing protocols across centers,” says Bancos. Rare Presentations, Important Lessons The two case reports, both published in JCEM Case Reports in March 2025, arrive at the same uncomfortable truth: that adrenal pathology can be diagnostically ambiguous. In “ Adrenocorticotropin-Secreting Pure Adrenal Ganglioneuroma Leading to Cushing Syndrome ,” corresponding author Daniel Alban, MD, of the Icahn School of Medicine at Mount Sinai in New York, N.Y. and team describe the second-only documented instance of an ACTH-producing pure adrenal ganglioneuroma (AGN) causing ACTH-dependent Cushing syndrome. (AGNs are typically benign, slow-growing, and hormonally inactive.) When a 23-year-old man presented to the team with hypertension, a right adrenal mass, anxiety and excessive sweating, and laboratory workup demonstrated ACTH-dependent hypercortisolism, with elevated 24-hour urinary free cortisol and failure to suppress on low-dose dexamethasone suppression testing, they suspected pheochromocytoma, a known precipitator of an ACTH-dependent hypercortisolemic state. However, normal plasma metanephrines and magnetic resonance imaging (MRI) of his pituitary undermined this working diagnosis without suggesting another cause of the hypercortisolism, prompting the team to pursue right adrenalectomy. Subsequent pathologic examination revealed a pure AGN, confirmed by positive ACTH immunostaining. This finding is remarkable not only for its rarity but also for what it suggests about microenvironmental influence on neoplastic behavior in that the authors hypothesize that the tumor’s location in the adrenal medulla may have exposed it to local corticotropin-releasing hormone, potentially stimulating ACTH production. Postoperatively, the patient’s blood pressure normalized without antihypertensive medication and his urinary free cortisol returned to normal. As this case demonstrates, AGNs present numerous diagnostic challenges: They are capable of demonstrating secretory properties with clinical presentations that vary based on the hormone(s) involved, secretory AGNs often mimic other adrenal pathologies (particularly pheochromocytomas), and definitive diagnosis requires histopathologic assessment. Moreover, it alerts clinicians to include ACTH-secreting AGN in the differential diagnosis when evaluating ACTH-dependent hypercortisolism in the presence of an adrenal lesion, even when the biochemical picture seems to point elsewhere. In “ Multiple Endocrine Neoplasia Type 1 With Adrenal Cortical Adrenocortical Carcinoma: A 25-Year Follow-Up and Family Report ,” Mei Yang, MD, and team of the Third People’s Hospital of Chengdu, in Chengdu, China, describe a 49-year-old woman who presented with recurrent hypoglycemia and was found on workup to have insulinoma, primary hyperparathyroidism, and a massive (>10 cm) left adrenal mass producing ACTH-independent hypercortisolism. Combined with her 25-year history of surgically treated pituitary macroprolactinoma, this constellation of symptoms pointed to MEN1, and genetic testing confirmed a heterozygous mutation in the MEN1 gene. On pathology, her adrenal tumor proved to be a rare and aggressive mucinous adrenocortical carcinoma. Despite surgery (adrenalectomy and distal pancreatectomy), the patient died from postoperative sepsis and septic shock, complications for which her concurrent Cushing syndrome increased her risk. Though sobering, this case has positive reverberations. Subsequent genetic testing of family members revealed that her 11-year-old son also carries the MEN1 mutation, which itself confers a high likelihood of future tumor development. The authors emphasize that this is precisely why genetic testing and family screening matter — not only for diagnosis, but also for the early detection that can change outcomes. Taken together, these four studies fill in important pieces to a puzzle that, for patients with adrenal disease, can have profound consequences when left unsolved. Horvath is a freelance writer based in Baltimore, Md. In the March issue, she wrote about how GLP-1s have been helping other conditions aside from diabetes and obesity. The post Piecing Together the Adrenal Puzzle appeared first on Endocrine News .