Libin Liu,1,* Chuyan Wang,2 Bin Mei,1 Jing Wang,1 Xiaoqun Xu,1 Hongjuan Zhou,1 Long Cai1,* 

1Centre of Laboratory Medicine, Hangzhou Red Cross Hospital, Hangzhou, Zhejiang, People’s Republic of China; 2Department of Clinical Laboratory, The Third People’s Hospital of Lin’an District, Hangzhou, Zhejiang, People’s Republic of China

*These authors contributed equally to this work

Correspondence: Long Cai, Centre of Laboratory Medicine, Hangzhou Red Cross Hospital, No. 208 East Huancheng Road, Hangzhou, 310003, People’s Republic of China, Email cailong317@hotmail.com

Purpose: This study aimed to establish and validate a diagnostic nomogram for identifying false positives in the Xpert MTB/RIF (Xpert) for detection of rifampicin resistance (RIF-R).
Patients and Methods: In this retrospective study, we collected basic patient characteristics and various clinical information from the electronic medical record database. Patients were randomly divided into training and validation groups in a 7:3 ratio. LASSO regression was used to screen variables and construct a diagnostic nomogram. The ROC curve, calibration curve, and decision curve analysis (DCA) were used to evaluate the performance of the nomogram.
Results: A total of 384 patients were included in the study, with 268 and 116 patients in the training and validation cohorts, respectively. Finally, probe mutations and probe delay were identified as the independent influencing factors. Using the mutation of probe E as a reference, probes A or C (OR = 51.07, P< 0.001), probe D (OR = 7.48, P< 0.001), and multiple probes (OR = 4.42, P=0.029) were identified as factors influencing false positives in Xpert for detection of RIF-R. Taking probe delay ΔCT < 4 as a reference, ΔCT (4– 5.9) (OR = 17.06, P=0.005) and ΔCT (6– 7.9) (OR = 36.67, P< 0.001) were noted to be the factors influencing false positives in Xpert for detection of RIF-R. Based on these two variables, we constructed a diagnostic nomogram. The area under the curve of the nomogram model was 0.847 and 0.850 for the training and validation groups, respectively. The calibration curves were consistent. The DCA revealed that the model achieved the greatest net benefit when the threshold probability was set between 6% and 71% in the training cohort and 6% and 70% in the validation cohort.
Conclusion: The nomogram constructed can identify false positives in Xpert for detection of RIF-R and provides basis for clinicians to formulate diagnosis and treatment plans.

Keywords: nomogram, probe mutation, probe delay, calibration curve