Endocrine Abstracts

Introduction: Distinguishing arginine vasopressin deficiency (AVP-D, central diabetes insipidus) from primary polydipsia (PP) is challenging. The method with the highest diagnostic accuracy, hypertonic saline-stimulated copeptin, is often limited to experienced hospitals, requires close monitoring, and may be cumbersome for patients. A standardised, validated stepwise assessment and probability score to rule out AVP-D in the initial baseline evaluation is currently lacking.

Methods: This analysis includes two independent patient cohorts from two international multicentre studies undergoing the hypertonic saline stimulation test (HST). Eligible adults with polyuria (>50 mL/kg body weight/day) or known AVP-D were included. The final diagnosis was based on medical history, clinical symptoms, laboratory/imaging data, results of the HST and therapeutic response at a three-month follow-up. Primary aim was to assess the diagnostic potential of baseline laboratory levels and to develop a novel probability score. In the probability score, the overall best cut-off, high-sensitivity cut-off (defined as ≥95% sensitivity), and high-specificity cut-off (defined as ≥95% specificity) were derived in the first cohort (development), and their diagnostic performance was determined in the second cohort (validation). The final scoring scheme included: (baseline plasma sodium x osmolality)/100), baseline plasma copeptin (–50 points, if >4.9pmol/l), nycturia [+30 points, if ≥3 times], onset of polyuria/polydipsia [+20 points, if sudden onset], drinking amount at night [30points, if >1L], presence of anterior pituitary deficiencies [+50points], previous pituitary surgery [+50points].

Results: In total, 299 patients were included between July 2013 and September 2022; 141 in the development cohort (n=59 [42%] AVP-D; n=82 [58%] PP) and 158 in the validation cohort (n=69 [44%] AVP-D; n=89 [56%] PP). For diagnosing AVP-D, a baseline plasma sodium of >145mmol/l provided 100% sensitivity (95%CI:100-100) and <135mmol 100% specificity (95%CI:100-100), and baseline plasma copeptin >5.6pmol/l resulted in 100% specificity (95%:CI 100-100). In the validation cohort, a probability score cut-off of 441 points provided the highest overall diagnostic accuracy of 86% (95%CI:73–94) and an area under the roc curve (AUC) of 91% (95%CI:87–96); the high-sensitivity cut-off of 415 points had a 93% sensitivity (95%CI:86-99), and the high-specificity cut-off of 461 points a 93% specificity (95%CI:88-98).

Interpretation: Our findings show that a plasma sodium <135mmol/l or a copeptin >5.6pmol/l rule out AVP-D, whereas a sodium >145mmoL/l can be used as a rule-in test. A novel probability score, including baseline laboratory values, clinical symptoms, and medical history, had high accuracy for identifying AVP-D without the need for any further stimulation testing.