This study was designed as a prospective randomized study conducted at the Chonnam National University Hospital. From August 1999 to July 2000, patients with carcinoma of the uterine cervix were enrolled in the study, and randomly assigned to receive ICR fractionation regimens of fraction size 3 Gy or 5 Gy. These two fraction sizes were selected for the following reasons. In Korea, the 5 Gy fraction is most commonly used, and the 3 Gy fraction is the smallest fraction used. The latter may have more potential to spare normal tissue radiobiologically, at the expense of more protracted fractionation. Before randomization, patients were stratified according to the International Federation of Gynecology and Obstetrics (FIGO) as stage Ib-IIa or stage IIb-IVa. Other stratification factors were not considered because of the relatively small study sample size.

Eligibility criteria consisted of a pathologically verified squamous cell carcinoma or adenocarcinoma of the cervix, and an Eastern Cooperative Oncology Group (ECOG) performance status index of 0-2. The patients had no history of previous malignancy, and hepatic, renal, and cardiopulmonary functions were adequate. Pretreatment evaluation included a complete medical history taking and a physical examination, and baseline blood tests, chest radiography, sigmoidoscopy, cystoscopy, and intravenous pyelography, and if necessary, computed tomography (CT) or magnetic resonance imaging (MRI) of the abdomino-pelvis. Informed consents were obtained from all patients, and the study protocol was approved by the institutional review board of Chonnam National University Hospital.

Patients in each stratum of stage Ib-IIa and stage IIb-IVa were assigned to either an ICR regimen of 3 Gy fractions (Group A) or 5 Gy fractions (Group B) by using a computer to randomly generate numbers. Prior to ICR, all patients underwent EBRT to the whole pelvis. EBRT was performed using a 10 MV linear accelerator using the 4-field box technique in daily fractions of 1.8 Gy, 5 days per week. EBRT prior to midline shield was delivered up to around 30.6 Gy over 3.5 weeks regardless of stage to allow a more controlled comparison of cumulative BED. After ICR started, further EBRT with midline shield was delivered on non-ICR weekdays up to total dose of 45 Gy, and if necessary, a booster dose of 5.4-14.4 Gy was delivered to the lateral pelvic wall.

Brachytherapy was performed by using a remotely controlled afterloading system containing a high-dose-rate radioisotope of ¹⁹²Ir (Varisource®, Varian, U.S.A.). A curved uterine tandem and paired ovoids were used as applicators. Applicator insertion was done on an outpatient basis with non-narcotic analgesics. For ICR simulation, orthogonal radiography films of anteroposterior and lateral view were taken with the applicators inserted, and the position of rectum, bladder, and the A point were defined according to ICRU 38 recommendations (14). The reference points of the rectum and bladder were identified on films by inserting a Foley catheter balloon filled with 5 mL of contrast media and by filling the rectum with 50 mL of barium. The rectal point dose was estimated at the anterior rectal wall on a line from the middle of the ovoid sources. ICR isodose curves were optimized by adjusting the dwell times of a 5 mm-length stepping source at each position for the three applicators. ICR was initially administered as a fraction size of 3 Gy to the A point, 3 times a week in 10 fractions (30 Gy) in group A, and 5 Gy twice a week in 5 fractions (25 Gy) in group B. This was planned as a similar total BED at A point for both groups. At this time, follow-up gynecological examinations or radiological studies, such as MRI, were undertaken to assess tumor shrinkage. According to the magnitude of the tumor regression and the resultant anatomical changes of the uterus, a secondary ICR boost was performed at a much smaller fraction size to A point as well as to the rectal and bladder reference point, with sufficient coverage of the residual tumor. This additional ICR boost was delivered 5 times in group A and 3 times in group B. The biological effective dose was calculated by using a linear-quadratic model with α/β ratio of 10 for tumors and an α/β ratio of 3 for the rectum and bladder. When calculating BED, we did not consider tumor repopulation during treatment. The equations used for the BED calculation at each reference point were as follows:

Chemotherapy was recommended in patients with adequate hepatic, renal, and cardiopulmonary functions, who were diagnosed as FIGO Ib-IIa with a bulky tumor (greater than 4 cm) or radiological pelvic lymphadenopathy, and in all patients with FIGO IIb or greater. The chemotherapy agents used were 5-FU (1,000 mg/day for day 1-5, intravenous infusion) and cisplatin (110 mg on day 1, intravenously). The timing of chemotherapy was planned to start at the same time as the external radiotherapy, and was repeated every 4 weeks for a total of four courses.

Each institution in Korea has different HDR equipment and treatment policies, which makes it difficult to conduct a multicenter study with an optimal sample size. Thus, we conducted as a single institutional pilot study and enrolled patients over a period of 12 months. The endpoints used to compare the effect of the two fractionation regimens were survival, pelvic tumor control, and complications. Toxicity was assessed using the Radiation Therapy Oncology Group (RTOG) grading system (15). Overall survival was defined as the observed length of life from the date of protocol entry to death or the last follow-up. Disease-specific survival was calculated from the time of entry to death due to disease or until the last follow-up. Patients who developed a central or a pelvic relapse were considered as pelvic failures. We assessed the tumor response 8 weeks after completing radiotherapy. Patients were followed up every 2 months over the first year, 3 monthly during the next 3 yr, and then 6 monthly during the subsequent years. Actuarial survival rates were calculated using the Kaplan-Meier method, and differences between survival curves were assessed using the log-rank test. Comparisons between categorical variables were based on the χ² test or Fisher's exact test. Comparisons of continuous variables between independent groups were performed using the Mann-Whitney U test. The Cox proportional hazards model was used to assess the importance of various covariables. The Statistical Package for the Social Sciences (SPSS) was used for statistical analysis.

A total of 56 patients were enrolled between August 1999 and July 2000 and each treatment group included 28 patients (Table 1). However, ten patients did not complete the planned treatment. Of these, seven patients refused treatment and three patients requested transfers to other hospitals. The resultant total number of eligible patients was 46 with 23 patients in each group. The median follow-up time was 42 months (range, 5-49 months) and the follow-up rate was 43/46 (93.5%) in all patients, 21/23 in group A and 22/23 in group B. The age range of all patients was 37-83 yr (median, 58 yr). The numbers of patients according to FIGO stage Ib₁, Ib₂, IIa, IIb, IIIa, and IIIb were 8 (17.4%), 3 (6.5%), 10 (21.7%), 22 (47.8%), 1 (2.2%), and 2 (4.3%), respectively. Detailed data on patient characteristics are listed in Table 2, and these characteristics were not significantly different in the two groups. Radiological lymphadenopathy (LAP) was defined as pelvic or paraaortic lymph node (PAN) enlargement by CT or MRI. Maximum tumor diameter was measured by CT or MRI. Detailed data on treatment characteristics are listed in Table 3. Radiotherapy duration and ICR number differed in the two groups because of the protracted fractionation in group A. Six patients were treated by ICR more than 15 times. In these patients, secondary or even tertiary ICR boost sessions with much smaller fractions were possible because rectal or bladder reference point doses were negligible. One patient who had paraaortic LAP was treated by extended field radiotherapy up to 45 Gy. Total physical dose at the A point including EBRT and ICR seemed to be larger in group A than in group B but the median total BED₁₀ at the A points were similar in the two groups.

The 3-yr overall survival rate (3YOSR) of all 46 patients was 76.1% (Fig. 1). According to FIGO stage, the 3YOSRs of Ib-IIa (n=21) and IIb-III (n=25) were 90.5% and 64.0%, respectively (p=0.04, Fig. 2), and the 3YOSRs of groups A and B were 78.3% and 73.9%, respectively (p=0.85, Fig. 3). There was no significant difference in 3YOSRs by age (<60 yr vs. ≥60), histology, tumor diameter (≤4 cm vs. >4), radiological LAP (pelvic vs. paraaortic vs. none), radiotherapy duration (≤60 days vs. >60), total A point BED₁₀ (≤110 Gy vs. >110), or chemotherapy (yes vs. no). A total of 11 patients died at the time of analysis, five patients from cervical cancer and the other six patients without evidence of cervical cancer. Thereby, the 3-yr disease-specific survival rate (3YDSSR) of all patients was 87.8%. The 3YDSSRs of groups A and B were 90.5% and 85.0%, respectively (p=0.71, Fig. 4). According to FIGO stage, the 3YDSSRs of Ib-IIa and IIb-III were 90.5% and 84.9%, respectively (p=0.64).

Forty-five of 46 patients (97.8%) achieved an initial complete response. Four patients (9.3%) developed pelvic failures among 43 patients who were evaluable for tumor control (Table 4). In group A (n=21), two patients developed combined pelvic and distant failures at 9 and 20 months, respectively; a 3-yr pelvic control rate of 90.0% (Fig. 5). In group B (n=22), one patient had local failure at 19 months and another patient developed combined failure at 18 months; a 3-yr pelvic control rate of 90.9%. Distant metastasis developed in 5 patients, who had all radiological lymphadenopathy. The sites of involvement were the paraaortic lymph nodes in 2, lung in 2, and one patient had both paraaortic and supraclavicular lymph node metastasis. The two groups were not significantly different in terms of the total failure rate (Table 4).

Acute toxicities, which included hematologic profiles, were evaluated mainly in patients who received concurrent chemotherapy. No patient showed grade 3 or 4 toxicity in terms of the hematologic, gastrointestinal, or genitourinary systems. All acute toxicities were relieved spontaneously or controlled with minor medications (grade 1 or 2, Table 5). Unlike other toxicities, anemia in our patients might reflect the effect of cervical cancer per se, rather than treatment. The two groups were similar in terms of the incidences of acute toxicities.

Excluding three patients, for whom information regarding late complications was unavailable, five of the 21 patients (23.8%) in group A, and two of 22 (9.1%) in group B developed late complications, but this difference was not statistically significant, because of the small sample size (p=0.240). In group A, all five patients developed grade 2 complications (Table 6). In group B, one patient developed grade 2 rectal bleeding and another patient combined rectovaginal and vesicovaginal fistulas of grade 4. The median delay for the occurrence of rectal complications was 16 months, whereas it was 21 months for bladder complications. The mean doses at various reference points were compared in patients with or without rectal complications of grade 2 or 4 in Table 7. As listed in Table 7, ICR A point dose (40.5 Gy vs. 34.9 Gy, p=0.026) and total A point BED₁₀ (119 Gy vs. 110 Gy, p=0.042) were significantly higher in patients with rectal complications. The mean doses at reference points were compared in patients with or without bladder complications of grade 2 or 4 in Table 8. As listed in Table 8, ICR A point dose (43.8 Gy vs. 35.1 Gy, p=0.022) and total bladder point BED₃ (144 Gy vs. 117 Gy, p=0.045) were significantly higher in patients with bladder complications.