Skip to main content

Table 3 Multinomial logistic regression analysis of various radiologic factors

From: CT and MRI features of sarcomatoid urothelial carcinoma of the bladder and its differential diagnosis with conventional urothelial carcinoma

Factors

Multinomial regression

Cutoff

AUC (95%CI)

Accuracy

Sensitivity

Specificity

PPV

NPV

 

Category

p

OR (95%CI)

       

Location

    

0.661(0.548–0.761)

0.651

0.639

0.682

0.848

0.405

 

Trigone

0.157

        

Shape

    

0.726(0.617–0.818)

0.795

0.869

0.591

0.855

0.619

 

Endophytic

0.487

        
 

Exophytic

0.614

        
 

Mixed

0.029

        

LAD

 

0.763

 

3.25

0.788(0.684–0.870)

0.675

0.590

0.909

0.947

0.444

SAD

 

0.014

1.053(1.011,1.098)

2.39

0.797(0.694–0.877)

0.699

0.639

0.864

0.929

0.463

EVE

    

0.702(0.591–0.797)

0.711

0.721

0.682

0.863

0.469

 

Yes

0.196

        

PPS

    

0.610(0.497–0.715)

0.747

0.902

0.318

0.786

0.539

 

Yes

0.204

        

Necrosis

    

0.732(0.623–0.823)

0.819

0.918

0.546

0.849

0.706

 

Yes

0.003

7.488(2.001,28.021)

       

Hydronephrosis/Ureteral effusion

 

0.367

  

0.637(0.524–0.740)

0.723

0.820

0.455

0.807

0.476

Combined model of SAD and Necrosis

    

0.849(0.754–0.919)

0.735

0.672

0.909

0.954

0.500

  1. Note—LAD, long-axis diameter; SAD, short-axis diameter; EVE, extravesical extension; PPS, pelvic peritoneal spread; OR, odds ratio; 95%CI, 95% confidence interval; AUC, area under the characteristic curve; 95%CI, 95% confidence interval; PPV, positive predictive value; NPV, negative predictive value