Research Article

Predicting Mismatch-Repair Status in Rectal Cancer Using Multiparametric MRI-Based Radiomics Models: A Preliminary Study

Table 1

Demographic and pathological data of the patients in both cohorts.

VariablesCH cohort valueRJ cohort value
dMMR ()pMMR ()dMMR ()pMMR ()

Gender (male/female)16/456/350.1177/436/181.000
Age (year)0.7750.881
BMI (kg/m2)0.6600.747
Histological type
 Adenocarcinoma13730.2387410.639
 Mucinous adenocarcinoma718413
Pathological T stage
 T1-27320.9894201.000
 T3-41359734
Pathological N stage
 N08350.8985250.959
 N1-21256629
Clinical M stage
 M04260.4343161.000
 M11665836
Tumor location
 Upper380.7002100.589
 Middle1153832
 Lower630112
Differentiation
 Well3140.2131100.330
 Moderate1446826
 Poor331218
Tumor deposit
 No13620.7867280.475
 Yes729426
Lymphovascular invasion
 No12470.4988290.408
 Yes844325
Perineural invasion
 No12630.4255300.540
 Yes828624
Tumor budding
 No11660.1246350.764
 Yes925519
KRAS
 Wild type12670.2236330.946
 Mutant type824521
NRAS
 Wild type13580.9158300.473
 Mutant type733324
BRAF
 Wild type11520.8617280.475
 Mutant type939426
CEA
 <5 ng/ml14520.2898320.619
 ≥5 ng/ml639322
CA19-9
 <37 U/ml15770.48010390.354
 ≥ 37 U/ml514115

CH cohort: Changhai Hospital, training and test sets; RJ cohort: Ruijin Hospital Luwan Branch, validation set; BMI: body mass index; dMMR: deficient mismatch repair; pMMR: proficient mismatch repair; CEA: carcinoembryonic antigen; CA19-9: carbohydrate antigen 19-9. Postoperative blood samples.