Endocrine Abstracts

Secondary hyperparathyroidism is highly prevalent in patients with end-stage chronic kidney disease (CKD). Optimal range of PTH level for this cohort of patients is still controversial. The aim of the study was to use a neural network algorithm to determine the optimal point value and confidence interval for the average PTH value for patients with end-stage chronic kidney disease to maintain bone metabolism.

The study included 190 patients with end-stage CKD receiving renal replacement therapy. Blood levels of parathyroid hormone, total calcium (Ca), phosphorus (P), levels of markers of bone metabolism: alkaline phosphatase (ALP), osteocalcin (OK), C-terminal telopeptides of type I collagen (CTx) were evaluated in patients. Bone mineral density (BMD) was determined by double x-ray absorptiometry. For clustering patients into groups, a neural network algorithm (auto-encoder) was used, consisting of an encoder and a decoder. Four clusters were obtained, one of which determined optimal indicators of BMD status, markers of bone metabolism, and also the best patient survival. The optimal value of PTH in terms of supporting bone metabolism and better survival rates of dialysis patients is in the range of 114–490 pg/ml. The point estimate of the average PTH value is 234 pg/ml, the variance is 188 pg/ml. Both a decrease and an increase in PTH can lead to a critical imbalance in bone metabolism and an adverse outcome, including loss of BMD and shortened life expectancy. In addition to the level of PTH, an adaptive increase in osteocalcin and CTx, and an age-associated decrease in BMD are important in determining the prognosis of bone tissue and patient survival. These factors must be taken into account when diagnosing secondary hyperparathyroidism and determining the tactics of its treatment and observation.