Table 1 Summary of clinicopathologic features evaluated in the different models to predict MMR deficiency

No. of variables

7

6

6

7

6

Sex

Female

-

-

-

-

Age (years)

≥60

<50

<50

<50

-

Tumor location

Proximal

Proximal

Proximal

Proximal

Proximal

Growth pattern

Expanding

-

-

-

Expanding

Dirty necrosis

Lack of

-

-

Lack of

-

Mucinous/signet-ring components

≥10%

≥50% (including medullary carcinoma)

Any component

Any component

Amount in %

Solid component

-

-

-

Amount in %

TIL

≥7 TIL/10 HPF

≥5 TIL/HPF (10 HPFs searched)

≥ 5TIL/HPF (10 HPF searched)

≥2 TIL/mean of 5 HPF

≥4TIL/HPF

Differentiation

-

Poorly differentiated

-

Well or poorly differentiated

-

Crohn-like reaction

-

≥4 nodules/LPF

-

≥3 nodules/section

≥3 nodules/LPF

Peritumoral lymphocytic reaction

-

-

Banding of lymphocytes beyond advancing edge

-

-

Increased stromal plasma cells

-

-

>25% plasma cells/stromal immune cells

-

-

Scoring system

No score, 7-factor index, cut-off ≥4

Score: cut-off ≥1

Score: cut-off ≥2.5; “Simplified PREDICT”: no score, ≥2 features present

Score: cut-off ≥1 or ≥1.5

Score: cut-off <0.8

Sensitivity

92.3% (4 features of 7)

93%

96.9% (score ≥2.5)

92% (score 1)

78.0%

Specificity

75.3% (4 features of 7)

55%

76.6% (score ≥2.5)

46% (score 1)

93.4%

Method applied for determination of MMR deficiency

IHC (4 markers), BRAF mutation

MSI (10 markers) and IHC (4 markers); validation study: MSI (5 markers) and IHC (MLH1, MSH2), BRAF mutation

MSI (5 markers); validation cohort only 1 marker if age ≥75 years

MSI (4 markers)

MSI (11 markers)