Computer simulations of lymph node metastasis for optimizing the pathologic examination of sentinel lymph nodes in patients with breast carcinoma

BACKGROUND. Many empiric protocols are used to detect metastases in sentine l lymph nodes (SLNs), but comparison of the efficacy of these methods is im practical because tissue is lost in processing, making reassessment with an other policy difficult. Consequently, performance indicators of this test a re largely unknown. DESIGN. The authors retrospectively examined 112 SLNs removed from 89 patie nts with breast carcinoma treated at the authors' institution and used the histologic data to devise a mathematic model of a SLN with Matlab modeling software. The authors simulated examination of this computer-generated (vir tual) lymph node according to several macroscopic and histologic sampling p rotocols and for each protocol assessed the probability of detecting microm etastases of specified sizes. The authors used published costing figures to estimate the cost of the policies. RESULTS. Direct comparison of 6 sectioning strategies currently in use by p athology laboratories showed the chances of detecting a 500-mum metastasis ranged from 20% to 75%. Four of the 6 protocols had a less than 30% chance of detecting metastases of this size. The detection rate of smaller metasta ses was poorer. Cost was not a good discriminator because some policies wer e more efficient than others. CONCLUSIONS. The detection of metastases is highly dependent on the methods used to look for them. The authors' simulations suggest that commonly used methods of examining lymph nodes have high false-negative rates, particula rly for small metastases. There is an urgent need for pathologists and clin icians to agree on the minimum size of SLN metastases that will be sought b y histology and set standard methods for examining these lymph nodes. Cance r 2000;89:2527-37. (C) 2000 American Cancer Society.