|
Advertisement |
|
|
||
 |
|
|||||
|
|||||
|
 |
||||||
|
||||||
|
||||||
|
© 2003 American Society for Clinical Oncology
Alternatives to Cystectomy in Muscle-Invasive Bladder CancerAlbert Einstein Cancer Center, Bronx, NY To the Editor: Rodel et al1 are to be commended for the exhaustive report of the long-term experience with organ preservation for bladder cancer using combined-modality therapy with chemotherapy and radiation therapy (RT). Further, they have gone on to estimate predictive factors for complete response (CR) at restaging and prognostic factors for development of distant metastases and for overall survival (OS). I would like to bring to your and the authors’ attention some errors or missing information in the statistical section. In Figure 1 (page 3,063), a clear and easy-to-read flow chart has been presented. However, the numbers do not tally. In the box for Restaging-Transurethral Resection (TUR)/Complete Responders/Local Relapse, the N value has been presented as 102. However, in the box below it, when the salvage treatment figure is presented, the numbers add up to 107 ([cystectomy, 42] + [TUR ± instillation, 49] + [other RT/surgery, 16] = 107). On page 3,064, when the authors present data on salvage treatment for "nonresponding and recurrent carcinoma," they mention that 83 (20%) of 415 patients underwent salvage cystectomy. I disagree with the figure used in the denominator. The numerator has been derived from salvage cystectomy number of 42 for the complete responders (n = 288) arm added to the salvage cystectomy number of 41 for the noncomplete responders (n = 110) arm. The denominator should ideally be 288 + 110 = 398. The 17 patients who did not undergo restaging TUR have been excluded from the numerator but have erroneously been included in the denominator. The reader has not been given any information about the latter 17 patients and how many, if any, underwent salvage cystectomy. The percentage should be presented as 83 of 398 or 21% rather than as 20%, a minor but important point from the statistical perspective.
The presentation of the predictive and prognostic factors in Tables 2, 3, and 4 is interesting in that the authors have rightfully used denominators of 398, 415, and 415, respectively, because rates of CR are not assessable in the 17 patients who did not undergo restaging TUR, but these very same patients are assessable for development of distant disease and OS. However, in Tables 2, 3, and 4, although the authors calculate the variables using univariate and multivariate models, they fail to mention whether there was any interaction between R (degree of residual tumor after TUR), T stage (tumor depth), and N stage (nodal involvement). They also fail to mention the degree of correlation between these variables. The risk of multicollinearity in a multiple regression model cannot be overemphasized. One previous publication implicated the T stage as being predictive for local recurrence only,2 whereas many others have done so for survival. It is also well established that higher T stage predicts for higher risk of nodal involvement. Intuitively, it can be presumed that the likelihood of T stage for predicting presence or absence of residual tumor (R in this study) and of positive lymph node status is high. In short, there is a fairly high risk of multicollinearity in this model, and those data have not been presented. If multicollinearity does exist, it has implications for reasons in the disparity of significance values of the R, T, and N stage in the two models. In the univariate model, all three independent variables were found to be significant for all three dependent variables: CR at restaging, development of distant metastases, and OS. The results on the multivariate analysis differ and are summarized in Table 1
The authors do not conceive a reason for the differences in the significance of each of these three factors in the multivariate model as compared with the univariate model. In a recently published article, using a multivariate analysis, the authors found that T and N stage and response to radiation did predict for overall survival.3 In situations of multicolinearity, the multiple regression model is often erroneous. It is also important to understand the steps involved in the calculation of the regression values because variables can change their significance level depending on the order in which they are brought into the model. Again, this information on the development of the model and the calculation steps will be quite helpful to the reader.
REFERENCES
1. Rodel C, Grabenbauer GG, Kuhn R, et al: Combined-modality treatment and selective organ preservation in invasive bladder cancer: Long-term results. J Clin Oncol 20:3061–3071, 2002 2. Greven KM, Spera JA, Solin LJ, et al: Local recurrence after cystectomy alone for bladder carcinoma. Cancer 69:2767–2770, 1992[Medline] 3. Scrimger RA, Murtha AD, Parliament MB, et al: Muscle-invasive transitional cell carcinoma of the urinary bladder: A population-based study of patterns of care and prognostic factors. Int J Radiat Oncol Biol Phys 51:23–30, 2001[Medline]
Related Article
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|
|||||||||||
|
|||||||||||
|
|
Copyright © 2003 by the American Society of Clinical Oncology, Online ISSN: 1527-7755. Print ISSN: 0732-183X
|
