The paper by Drs. Moul and Heidenreich provides a very nice review of prognostic factors for metastasis in patients with clinical stage I nonseminoma. Risk-adapted management--ie, the management of patients at low risk for metastasis by surveillance and patients at high risk for metastasis by retroperitoneal lymph node dissection (RPLND)--is very reasonable, and we are now at a point where a paradigm can be developed to accurately classify clinical stage I patients as either low or high risk and manage them accordingly.
The paper describes nerve-sparing RPLND as a "major abdominal procedure." Although post-chemotherapy RPLND is associated with significant morbidity, nerve-sparing RPLND is not.[1] I also disagree with the authors' statement that vascular invasion and percentage of embryonal carcinoma have "extreme clinical utility," since these factors have not been verified prospectively in a consecutive group of patients. As is well known, retrospective studies in selected patients do not necessarily generate a paradigm that is proven to be clinically valid prospectively in unselected patients.
The authors' analysis of histologic and molecular factors is very well done and well referenced. However, they should have included in the clinical prognostic factors a section on lymph node diameter on CT scan as a predictor of pathologic stage.[2] The experience from Indiana would suggest that pathologic stage varies continuously with measured lymph node diameter on CT scan.
Need for a Prospective, Multi-institutional Study
Many of the studies described by Moul and Heidenreich were carried out retrospectively in selected patients. Furthermore, the paper accurately describes the potential problems with interobserver variability relative to histologic and immunohistochemical parameters. What is needed is a multi-institutional study of these parameters that is done prospectively in unselected patients. Such a study has the potential to validate one or more of the paradigms described in this paper and would allow us to determine whether or not interobserver variability can be minimized. Hopefully, such a study will be undertaken by the cooperative groups
