level of 0.05, the probability of finding one false-positive result increases with the number of tests performed. This probability increases from 0.098 with two tests, to 0.226 with 5 tests, and 0.460 with 12 tests. This problem, recognized by most statisticians, can be rectified by the Bonferroni correction (7), which calls for the desired overall significance level to be divided by the number of tests performed. To preserve an overall significance of 0.05 with the 12 pairwise tests of the prevalence rates here, the p value would have to be less than the adjusted value of 0.0042 (=0.05/12). We calculated all 12 p values and found the lowest to be 0.018, which is well above the adjusted critical value for statistical significance. More convincingly, however, Table 1 also shows no increase in the prevalence rates with HFX after 2.0 years and there is no statistical difference in the incidence rates computed using two different methodologies as presented in Tables 2 and 3. Finally, Dr. Logie recommends exploration of a combination of chemotherapy and radiation as a strategy specifically designed to avoid the late effects typical of radiation while maintaining a superior tumor cell kill compared to standard fractionation. In principle, it does make much more sense to design combined modality regimens based on specific mechanisms of radiation– drug interaction. In reality, however, knowledge on the exact mechanisms of drug–radiation interaction or the pharmacodynamics is limited for most cytotoxic agents. Conse- quently, most combined chemoradiation regimens evolved empirically simply by adding chemotherapy regimens showing some clinical activ- ity, usually in patients with disseminated disease, to radiation. There- fore, most of the concurrent chemoradiation regimens turn out to have increased acute or late toxicity relative to radiation alone though some also yielded survival advantage (8). The data of a recently completed Intergroup Phase III trial enrolling patients with unresectable head-and- neck squamous cell carcinoma, presented at the 2000 Annual Meeting of the American Society of Clinical Oncology, clearly illustrate this point (9). This trial randomized patients to radiation alone (70 Gy in 7 weeks), radiation (70 Gy) plus cisplatin (100 mg/m 2 for 3 cycles), or split-course radiation (30 Gy + 30 – 40 Gy) given with the first and third cycles of cisplatin (75 mg/m 2 ) plus fluorouracil (1000 mg/m 2 ) given every 4 weeks. The 2- and 3-year actuarial survival rates were 30 –20% for radiation alone, 43–37% for radiation + cisplatin ( p = 0.016), and 40 –29% for split-course radiation plus cisplatin–fluorouracil ( p = 0.13). The Grade 3+ toxicity, however, occurred in 53%, 86% ( p  0.0001), and 77% ( p  0.001) of the patients, respectively. Fortunately, advances in understanding the biology of head-and-neck cancers have brought us a step forward in identifying specific molecular targets for therapeutic inter- ventions, which hopefully will translate into a substantial therapeutic gain. K. KIAN ANG, M.D., PH.D. Department of Radiation Oncology University of Texas M. D. Anderson Cancer Center Houston, TX BRIAN A. BERKEY, M.S. THOMAS F. PAJAK,PH.D. RTOG Statistical Unit Philadelphia, PA KAREN K. FU, M.D. Department of Radiation Oncology University of California San Francisco San Francisco, CA PII S0360-3016(01)02777-8 1. Dubben HH, Beck-Bornholdt HP. A Radiation Therapy Oncology Group (RTOG) Phase III randomized study to compare hyperfraction- ation and two variants of accelerated fractionation to standard-frac- tionation radiotherapy for head-and-neck squamous cell carcinomas: First report of RTOG 9003: In regard to Fu et al. IJROBP 2000;48: 7–16. Actuarial estimates of late normal-tissue effects . . . now! Int J Radiat Oncol Biol Phys 2001;51:563. 2. Logie MB. Regarding Fu et al., A Radiation Therapy Oncology Group (RTOG) phase III randomized study to compare hyperfractionation and two variants of accelerated fractionation to standard fractionation radiotherapy for head and neck squamous cell carcinomas: First report of RTOG 9003. Int J Radiat Oncol Biol Phys. In press. 3. Fu KK, Pajak TF, Trotti A, et al. A Radiation Therapy Oncology Group (RTOG) phase III randomized study to compare hyperfraction- ation and two variants of accelerated fractionation to standard frac- tionation radiotherapy for head and neck squamous cell carcinomas: First report of RTOG 9003. Int J Radiat Oncol Biol Phys 2000;48:7– 16. 4. Kaplan EL, Meier P. Nonparametric estimation from incomplete ob- servations. J Am Stat Assoc 1958;53:457– 481. 5. Kalbfleish J, Prentice R. The statistical analysis of failure time data. New York: John Wiley and Sons; 1980. 6. Baumann M, Bentzen SM, Ang KK. Hyperfractionated radiotherapy in head and neck cancer: A second look at the clinical data. Radiother Oncol 1998;46:127–130. 7. Bonferroni CE. Teoria statistica delle classi e calcolo delle probabilita. Pubblicazioni del R Istituto Superiore di Scienze Economiche e Commerciali di Firenze 1936;8:3– 62. 8. El-Sayed S, Nelson N. Adjuvant and adjunctive chemotherapy in the management of squamous cell carcinoma of the head and neck region: A meta-analysis of prospective and randomised trials. J Clin Oncol 1996;14:838 – 847. 9. Adelstein DJ, Adams GL, Li Y, et al. A phase III comparison of standard radiation therapy (RT) versus RT plus concurrent cisplatin (DDP) versus split-course RT plus concurrent DDP and 5-fluorouracil (5FU) in patients with unresectable squamous cell head and neck cancer (SCHNC): An intergroup study. (Abstract). Proceedings of ASCO 2000;19:411a. 10. Mantel N. Evaluation of survival data and two new rank order statistics arising in its consideration. Cancer Chemother Rep 1966;5:163–170. 11. Gray RJ. A class of K-sample tests for comparing the cumulative incidence of a competing risk. Ann Stat 1988;16:1141–1154. ON THE CONSISTENT USE OF ORGAN DEFINITIONS AND RADIOBIOLOGICAL MODELS FOR THE EVALUATION OF COMPLICATION PROBABILITIES OF “TUBULAR” ORGANS To the Editor: The advent of 3D conformal therapy has instigated an enormous interest in the application of radiobiological models for the purposes of treatment planning optimization and evaluation, mostly be- cause of the potential of these models to summarize an abundance of dosimetric information into a single figure of merit (1). This figure of merit (whether tumor control probability [TCP] or normal tissue complication probability [NTCP]) depends on the functional form of the radiobiological model, organ definitions, planned dose distribution, and the values of the model free parameters. Figure 1a illustrates a parameter estimation procedure to emphasize that Table 3. Time to first Grade 3+ toxicity at various times after commencement of radiation (by Cumulative Incidence method) SFX HFX AFX-S AFX-C %  SE At risk %  SE At risk %  SE At risk %  SE At risk 1 year 17.2  2.5 146 16.9  2.5 144 16.0  2.4 149 17.6  2.5 149 2 years 22.2  2.8 77 23.3  2.8 90 21.6  2.7 84 25.2  2.9 90 3 years 25.6  3.0 49 26.5  3.0 58 24.9  2.9 50 29.0  3.1 58 4 years 27.1  3.1 26 26.5  3.0 35 27.7  3.1 29 31.6  3.2 38 5 years 27.1  3.1 17 26.5  3.0 24 28.5  3.1 21 34.9  3.6 18 Comparison with SFX (11) — p = 0.99 p = 0.94 p = 0.27 1150 I. J. Radiation Oncology ● Biology ● Physics Volume 52, Number 4, 2002