Naoki Yamada, Kumiko Ishigaki, Takeshi Hayakawa, Kaito Iida, Shoko Yamaoka, Pinhui Wang, Kazushi Asano
PRIs were derived for all elements. The MBI showed high apparent discrimination between PAC and controls (AUC, 0.983). At a cutoff of 0.8030, sensitivity, specificity, and accuracy were 100.0, 92.3, and 93.9%, respectively. Apparent AUCs were 0.984 with age-restricted controls and 0.998 for BPH versus PAC. As these were unvalidated training-set estimates, they should not be interpreted as clinical performance. In dogs with PAC, preoperative MBI was higher with lymph node metastasis, but did not differ by surgical margin status or clearly change 2 weeks postoperatively.
BACKGROUND: Noninvasive biomarkers for canine prostatic adenocarcinoma (PAC) are lacking. Serum multi-element profiling by inductively coupled plasma-mass spectrometry (ICP-MS) may provide diagnostic information.
OBJECTIVES: To derive preliminary reference intervals (PRIs) for 18 serum elements in clinically healthy male dogs, assess a Metallobalance index (MBI) for distinguishing PAC from controls, and explore associations with lymph node metastasis, surgical margin status, and postoperative change.
ANIMALS: The study included 115 dogs: clinically healthy males (n = 65), dogs with benign prostatic hyperplasia (BPH; n = 25), one dog with a prostatic cyst, and dogs with histopathologically confirmed PAC (n = 24). The diagnostic control group comprised 91 dogs.
METHODS: Serum concentrations of 18 elements were measured by ICP-MS, and central 95% PRIs were derived. An 18-element linear discriminant analysis (LDA) model generated the MBI. Cutoffs maximized Youden's index; apparent performance was summarized by sensitivity, specificity, accuracy, and area under the receiver operating characteristic curve (AUC). Post hoc analyses restricted controls to dogs aged ≥7 years and compared BPH with PAC using the Study 1 MBI equation without refitting. Element-specific and clinicopathological subgroup analyses were exploratory.
RESULTS: PRIs were derived for all elements. The MBI showed high apparent discrimination between PAC and controls (AUC, 0.983). At a cutoff of 0.8030, sensitivity, specificity, and accuracy were 100.0, 92.3, and 93.9%, respectively. Apparent AUCs were 0.984 with age-restricted controls and 0.998 for BPH versus PAC. As these were unvalidated training-set estimates, they should not be interpreted as clinical performance. In dogs with PAC, preoperative MBI was higher with lymph node metastasis, but did not differ by surgical margin status or clearly change 2 weeks postoperatively.
CONCLUSIONS AND CLINICAL RELEVANCE: ICP-MS-based multi-element profiling with LDA showed high apparent training-set discrimination of canine PAC from controls, including benign prostatic disease. The association with lymph node metastasis warrants further study, whereas findings did not support short-term postoperative monitoring or a validated prognostic role. Independent validation in a larger prospective cohort is required before clinical application.