Prediction of DNA mismatch repair gene mutation status in incident colorectal cancer cases*.

Colon Cancer Genetics Group
University of Edinburgh and
MRC Human genetics Unit, Edinburgh

Pr / (1-Pr) = 1.39 * 0.89Age * 2.57Gender * 4.45Location * 9.53Sync/MetTumour * 46.26CRCFH(<50) * 7.04CRCFH(>=50) * 59.36ECFH

Where Pr is the probability of a mutation carrier
Criteria are as follows:

Calculation

 
Is this a real patient?
Has mutation analysis been carried out?
Has the mutation been identified?
Mutation details:  
Please enter your email address:
Please enter the patient's age
Please enter the patient's gender
Please enter the tumours location
Sync/MetTumours
CRCFH
ECFH
 

*The model was developed from a population <55yrs and validated on a population <45yrs. We are unclear as to how it will perform in patient populations >55yrs.

Refinement of carrier prediction using clinical model with tumour analysis

Having estimated the carrier probability from clinical parameters, this can be refined using immunohistochemical staining (IHC) of tumour biopsy slides (diagnostic or resection specimen) or by later assessment of microsatellite instability (MSI) status. To guide these decisions, the table below shows sensitivity, specificity and positive predictive value (PPV) for Stage 1, and Stage 2 (tumour MSI /IHC). Negative predictive value is not shown because it is never less than 97% after the second stage with any model cut-off. The model cut-off relates to the stringency of the prediction and a cut-off probability of x means a person is categorised as mutation carrier if his/her predicted probability of being mutation carrier is >=x (non-carrier <x).

 

After Stage 1

After Stage 2

 
 Prospective Series Replication Series  MSI and IHC tumour analysis Identified by Stage 2
(% identified in Stage 1)
Cut-off Sensitivity Specificity Positive predice value Sensitivity Specificity Positive predice value Number passing to stage 2 (%) Sensitivity Specificity Positive predice value  
  MSI IHC MSI IHC MSI IHC MSI IHC
0.005 0.95
(0.82-0.99)
0.14
(0.12-0.17)
0.05
(0.03-0.07)
1.00 0.00 0.23 728 (86%) 0.88
(0.78-0.98)
0.81
(0.68-0.93)
0.87
(0.85-0.89)
0.94
(0.92-0.95)
0.24
(0.17-0.31)
0.38
(0.27-0.48)
138 (16%) 81 (9.3%)
0.01 0.89
(0.80-0.99)
0.42
(0.39-0.45)
0.07
(0.05-0.09)
0.94 0.075 0.23 506 (60%) 0.83
(0.71-0.95)
0.76
(0.62-0.89)
0.91
(0.89-0.93)
0.95
(0.94-0.97)
0.30
(0.22-0.39)
0.42
(0.31-0.54)
104 (12%) 68 (7.8%)
0.05 0.68
(0.51-0.82)
0.86
(0.83-0.88)
0.19
(0.13-0.26)
0.94 0.27 0.27 140 (17%) 0.65
(0.50-0.80)
0.62
(0.46-0.77)
0.97
(0.96-0.98)
0.99
(0.98-1.00)
0.53
(0.39-0.68)
0.80
(0.66-0.95)
46 (5.2%) 30 (3.4%)
0.15 0.55
(0.39-0.71)
0.96
(0.95-0.97)
0.40
(0.26-0.53)
0.86 0.64 0.41 53 (6%) 0.52
(0.36-0.68)
0.49
(0.33-0.65)
1.00
(0.99-1.00)
1.00
(1.00-1.00)
0.87
(0.73-1.00)
1.00
(0.85-1.00)
23 (2.6%) 19 (2.2%)
0.2 0.47
(0.31-0.63)
0.97
(0.96-0.98)
0.45
(0.30-0.60)
0.80 0.73 0.46 40 (5%) 0.44
(0.29-0.60)
0.44
(0.28-0.60)
1.00
(0.99-1.00)
1.00
(1.00-1.00)
0.8
(0.71-1.00)
1.00
(0.84-1.00)
20 (2.3%) 17 (2.0%)
0.25 0.39
(0.24-0.55)
0.98
(0.97-0.99)
0.45
(0.28-0.62)
0.74 0.76 0.47 33 (4%) 0.36
(0.21-0.51)
0.36
(0.21-0.51)
1.00
(0.99-1.00)
1.00
(1.00-1.00)
0.84
(0.67-1.00)
1.00
(0.81-1.00)
16 (1.8%) 14 (1.6%)
0.35 0.39
(0.24-0.55)
0.99
(0.98-1.00)
0.71
(0.52-0.91)
0.74 0.80 0.52 21 (2%) 0.36
(0.21-0.51)
0.36
(0.21-0.51)
1.00
(0.99-1.00)
1.00
(1.00-1.00)
0.87
(0.71-1.00)
1.00
(0.81-1.00)
16 (1.8%) 14 (1.4%)
0.45 0.34
(0.19-0.49)
1.00
(1.00-1.00)
1.00
(0.79-0.91)
0.66 0.85 0.56 13 (2%) NA NA NA NA NA NA NA NA
Criteria  
Bethesda 0.95
(0.82-0.99)
0.38
(0.34-0.41)
0.06
(0.05-0.09)
- - - 555 (64%) 0.88 (0.77-0.98) 0.84 (0.73-0.96) 0.92 (0.90-0.94) 0.95 (0.94-0.97) 0.32 (0.23-0.41) 0.44 (0.33-0.56) 103 (12%) 72 (8.3%)
AMS 0.42
(0.26-0.59)
0.98
(0.97-0.99)
0.47
(0.30-0.65)
- - - 34 (4%) 0.39 (0.24-0.55) 0.39 (0.24-0.54) 1.00 (1.00-1.00) 1.00 (1.00-1.00) 1.00 (0.82-1.00) 1.00 (0.82-1.00) 15 (1.7%) 15 (1.7%)

Notes

MSI status is assessed using MSS and either MSI-H or MSI-L categories because both had similar predictive value in univariate analysis. Hence, MSI is dichotomised to either stable (MSS) or unstable (MSI-H or MSI-L).. "Negative" immunohistochemistry refers to loss of expression of any gene, not necessarily the relevant gene for the germline mutation. Univariate analysis shows MLH1, MSH2 and MSH6 loss are individual significant predictors of mutation status, but there is a strong relationship between loss of expression of MSH2 and MSH6, reflecting the biochemical interaction of the proteins. Hence, adding MSH6 data to MLH1 and MSH2 in the model does not significantly better explain the data. "Any gene loss" is also useful in practice since it takes account of poor quality immunohistochemistry that might be obtained for one antibody but good quality slides for another showing clear loss of expression.

To comment on this research or the calculation please contact Professor Dunlop