Postoperative MSCC overall mortality calculator

Predicts the overall mortality for (postoperative) patients with maxillary squamous cell carcinoma with clinical and histopathological predictors.
How to use

This prediction model includes both clinical predictors and histopathological predictors, whilst adjusting for the confounding effect of age, medical history and comorbidities with Charlson’s Comorbidity Index (CCI).


The CCI is a validated index that calculates how age, medical history and comorbidities affect 10-year overall survival of patients. The CCI can be scored easily with this calculator or at MDCalc.


After tumour-staging and treatment planning, this prediction model can be used to determine the effects of (in)complete tumour resection margins and the necessity for neck treatment, on 2- and 5-year overall mortality, whilst adjusting for CCI.


Moreover, should postoperative histopathological analysis determine that the tumour was positive for bone invasion, spindle growth and/or vasoinvasive growth, the 2- and 5-year mortality probabilities can be recalculated with inclusion of these predictors.

Select relevent predictors

Close/free surgical margins (>1mm)
0
Positive margins (<1mm)
1
No neck treatment
0
Neck dissection, radiotherapy of the neck, and/or sentinel node
1
Absent
0
Present
1
Absent
0
Present
1
Absent
0
Present
1
points
Calculate Charlson's comorbidity index
<50 years
0
50–59 years
+1
60–69 years
+2
70–79 years
+3
≥80 years
+4
No
0
Yes
+1
No
0
Yes
+1
No
0
Yes
+1
None or diet-controlled
0
Uncomplicated
+1
End-organ damage
+2
No
0
Yes
+2
None
0
Localized
+2
Metastatic
+6
No
0
Yes
+2
No
0
Yes
+2
No
0
Yes
+6
No
0
Yes
+2
No
0
Yes
+1
No
0
Yes
+1
No
0
Yes
+1
No
0
Yes
+1
No
0
Yes
+1
None
0
Mild
+1
Moderate to severe
+3


Result:

Please fill out required fields.
About the prediction models
Summary of predictors
Creators

Background

Development

Potential predictors were identified based on a literary search and expert opinion. Predictor selection was conducted with stepwise backward selection, with elimination if Akaike’s criterion (p<.157) was met. In this study, both Cox regression and Logistic regression were used to develop models for the prediction 2- and 5-year mortality. Only the logistic regression models were converted into the probability equations, because they are easy to use and interpret.

Validation

Both the 2- and 5-year models were internally validated with bootstrapping with 100 resamples with replacement and adjusted for overfitting by recalibration with a shrinkage factor.

No external validation has yet been performed.

The performance test results are listed in this table:

Omnibus
Tests of
Model
Coefficients		 Hosmer and
Lemeshow
test		 Variance
Nagelkerke’s
R2 (%)		 Sensitivity (%) Specificity (%) PPV (%) NPV (%) Accuracy (%) C-index C-index bootstrap
2-year mortality adjusted Χ2(6) =30.851,
p<.0001		 Χ2(7) =1.261,
p=.989		 40.1% 61.5% 87.0% 64.0% 85.7% 80.0% .841 .838
5-year mortality adjusted Χ2(5) =21.698,
p=.001		 Χ2(7) =4.317,
p=.743		 30.5% 58.8% 88.2% 76.9% 76.3% 76.5% .770 .749

Summary of predictors

Surgical margins 0 = close/free surgical margins (>1mm) 1 = positive margins (<1mm)
Treatment of the neck 0 = surveillance, no neck treatment 1 = neck dissection, radiotherapy of the neck, and/or sentinel node
Spindle growth 0 = absent 1 = present
Vasoinvasive growth 0 = absent 1 = present
Bone invasion 0 = absent 1 = present
Charlson’s Comorbidity Index (CCI) points, n= … (see here)

2-year mortality equation

$$P = {e^{\ 0.805\ \times \ (((neck\ treatment\ \times \ 1.665) \ \ + \ \ (surgical\ margins\ \times \ 1.008)\ \ + \ \ (bone\ inv.\ \times \ 1.309)\ \ + \ \ (spindle\ gr.\ \times \ 1.500))\ \ + \ \ (vasoinv.\ gr.\ \times \ 1.559))\ \ + \ \ (CCI\ \times \ 0.099))\ - \ 4.008)} \over 1 + e^{\ 0.805\ \times \ (((neck\ treatment\ \times \ 1.665) \ \ + \ \ (surgical\ margins\ \times \ 1.008)\ \ + \ \ (bone\ inv.\ \times \ 1.309)\ \ + \ \ (spindle\ gr.\ \times \ 1.500))\ \ + \ \ (vasoinv.\ gr.\ \times \ 1.559))\ \ + \ \ (CCI\ \times \ 0.099))\ - \ 4.008)}}$$

5-year mortality equation

$$P = {e^{\ 0.770\ \times \ (((neck\ treatment\ \times \ 1.375) \ \ + \ \ (surgical\ margins\ \times \ 0.996)\ \ + \ \ (spindle\ gr.\ \times \ 1.307))\ \ + \ \ (vasoinvasive\ gr.\ \times \ 1.629))\ \ + \ \ (CCI\ \times \ 0.093))\ - \ 2.108)} \over 1 + e^{\ 0.770\ \times \ (((neck\ treatment\ \times \ 1.375) \ \ + \ \ (surgical\ margins\ \times \ 0.996)\ \ + \ \ (spindle\ gr.\ \times \ 1.307))\ \ + \ \ (vasoinvasive\ gr.\ \times \ 1.629))\ \ + \ \ (CCI\ \times \ 0.093))\ - \ 2.108)}}$$

Fons Slieker, MD, Msc

About the Creator

Fons Slieker, MD, is a PhD-Candidate in Head and Neck Oncology at the University Medical Center Utrecht. His thesis focuses on maxillary carcinoma.

To view Fons Slieker ' s publications, visit PubMed


Web calculator created by Y.R. Aerssens, MD, Msc. Design based of mdcalc.com