How to Compute the Probability of Nodal Involvement in Invasive Breast Cancer

Step-by-Step Guide

1. 1

   Step 1: Obtain the probability of nodal involvement by reference to factors such as clinical palpability

   Useful tables can be found in the medical literature.
   
   Call this probability the prior probability.
   
   See Table 6 in and Table 1 below which is a modification of Table 6 of by Professor Howell Tong (London School of Economics), see for details.
   
   Table 1:
   APPROXIMATE NUMBER OF PATIENTS PER 100 PREDICTED TO HAVE AXILLARY NODE INVOLVEMENT BEFORE AXILLARY SAMPLING (LVI = lymphovascular invasion) Tumour Size (in mm.) 0~5 6~10 11~15 16~20 21~25 26~30 31~50 51~100 LVI absent Can you feel the tumour? No 4 11 12 10 20 33 57 67 Yes 6 13 19 20 30 27 34 65 Yes; axillary nodes too 50 75 93 80 70 71 77 93 LVI present Can you feel the tumour? No 5 18 30 39 42 33 50 95 Yes 23 28 37 49 65 67 66 77 Yes; axillary nodes too 98 98 98 92 90 96 98 98
2. 2

   Step 2: tumour size

   We may use the following values which are based on the conservative end of the range given in Table 1 of .
   
   Number of nodes, n 3 4 5 6 Probability of false negative, FN(n)
   0.11
   0.06
   0.05
   0.03 , Then apply the following formula:
   Posterior probability with a sample of n nodes which are all clear = FN(n) x prior probability / (1
   - prior probability + FN(n) x prior probability).
   
   For small FN(n) x prior probability, which is usually the case, the posterior probability is approximately equally to FN(n) x prior probability / (1
   - prior probability), i.e.
   
   False negative probability x odds (of nodal involvement).
   
   Alternatively, use Table 1 to obtain a prediction before axillary sampling and then Table 2 to obtain your prediction after axillary sampling.
   
   Table 2 does not apply to samples that produce one or more positive nodes.
   
   Example:
   Suppose the patient's breast tumour has size
   1.6 centimeter (0.6 in). with no lymphovascular invasion and she discovered it by her feeling the lump herself.
   
   Reading off from Table 1, the prior probability of nodal involvement is equal to
   0.2.
   
   Suppose 5 axillary nodes have been removed from her by the surgeon and all the nodes are found to be clear.
   
   Then because FN(5)=0.05, the posterior probability of nodal involvement =
   0.05 x
   0.2 / (1-0.2 +
   0.05 x
   0.2 ) =
   0.0123=1.23%.
   
   The approximate formula gives
   1.25%. , Table 2:
   APPROXIMATE NUMBER OF PATIENTS PER 100 PREDICTED TO HAVE AXILLARY NODE INVOLVEMENT AFTER AXILLARY SAMPLING Number of nodes sampled and found to be clear 3 4 5 6 Number of patients per 100 having axillary node involvementbefore sample taking 5 1 0 0 0 10 1 1 1 0 15 2 1 1 1 20 3 2 1 1 25 4 2 2 1 30 5 3 2 1 35 6 4 3 2 40 7 4 3 2 45 8 5 4 2 50 10 7 5 3 55 12 8 6 4 60 14 10 7 4 65 17 12 8 5 70 20 14 10 7 75 25 17 13 8 80 31 22 17 11 85 38 28 22 15 90 50 39 31 21 95 68 57 49 36 100 100 100 100 100 Table 2 was constructed by Professor Howell Tong (London School of Economics), see for details.
3. 3

   Step 3: lymphovascular invasion etc.

4. 4

   Step 4: Obtain the probability of false negative pertaining to a sample of n nodes.

5. 5

   Step 5: Call the desired probability of nodal involvement on the basis of a pathologic analysis of the sampled nodes the posterior probability.

6. 6

   Step 6: Alternatively

7. 7

   Step 7: read the posterior probability from Table 2 below

8. 8

   Step 8: which is found to be approximately 1% for a woman whose prior probability equals 20% and all 5 nodes sampled from her are found to be clear.

Detailed Guide

About the Author