Late rectal complication (LRC) was a major late complication in patients with uterine cervical carcinoma who were treated with a combination of external beam radiotherapy (EBRT) and high-dose-rate intracavitary irradiation (HDR-ICR). For the current study, the authors retrospectively evaluated dosimetric parameters that were correlated with LRC ≥ Grade 2 in patients with uterine cervical carcinoma who were treated with curative radiotherapy, and they analyzed the appropriate dose estimates to the rectum that were predictive for LRC ≥ Grade 2.
Between July 1994 and September 2002, 157 patients who were diagnosed with Stage IB–IIIB cervical carcinoma and were treated with definitive radiotherapy were included. EBRT (41.4–66 grays [Gy] in 23–33 fractions) to the whole pelvis was delivered to all patients, with midline shielding performed after a 36–50.4 Gy external dose. HDR-ICR (21–39 Gy in 6–13 fractions to Point A) was administered at a rate of 2 fractions weekly after midline shielding of EBRT. LRC was scored using Radiation Therapy Oncology Group criteria. The total biologically effective dose (BED) at specific points, such as Point A (BEDPoint A), rectal point (BEDRP), and maximal rectal point (BEDMP), was determined by a summation of the EBRT and HDR-ICR components, in which the α/β ratio was set to 3. Analyzed parameters included patient age, tumor size, stage, concurrent chemotherapy, ICR fraction size, RP ratio (dose at the rectal point according to the Point A dose), MP ratio (dose at the maximal rectal point according to the Point A dose), EBRT dose, BEDPoint A, BEDRP, and BEDMP.
The 5-year actuarial overall rate of LRC ≥ Grade 2 in all patients was 18.4%. Univariate analysis showed that the RP ratio, MP ratio, EBRT dose, BEDPoint A, BEDRP, and BEDMP were correlated with LRC ≥ Grade 2 (P < 0.05). Multivariate analysis showed that, of all clinical and dosimetric parameters evaluated, only BEDRP was correlated with LRC ≥ Grade 2 (P = 0.009). The 5-year actuarial rate of LRC ≥ Grade 2 was 5.4% in patients with a BEDRP < 125 Gy3 and 36.1% in patients with a BEDRP ≥ 125 Gy3 (P < 0.001).
Radiotherapy (RT) plays an important role in the treatment of patients with carcinoma of the uterine cervix. Definitive RT usually involves a combination of external beam radiation therapy (EBRT) to the whole pelvis and intracavitary irradiation (ICR). High-dose-rate ICR (HDR-ICR) is used widely, because the local control rate and the incidence of late complications with HDR-ICR are comparable to those for low-dose-rate ICR (LDR-ICR).1, 2 When analyzing treatment outcomes for patients with uterine cervical carcinoma, it is important to assess not only tumor control but also the incidence of late complications. Among the late complications, the most frequent and major late complications after definitive RT for uterine cervical carcinoma are rectal complications, such as proctitis, bleeding, and fistulas.
Several investigators3–17 previously demonstrated that various clinical and treatment-related parameters, including age, diabetes mellitus, disease stage, EBRT dose, fractionation of brachytherapy, Point A dose, and rectal dose, were correlated with late rectal complications (LRC). In particular, RT dose and volume are considered the most important parameters associated with LRC.10, 12, 14–16 However, the analysis for the risk of LRC is somewhat difficult due to the highly heterogeneous dose distribution within the pelvis, the variations in combined EBRT and HDR-ICR therapy, and the complexity of anatomy. Therefore, the reliable estimates of doses to the rectum for predicting the risk of LRC in patients with uterine cervical carcinoma who are treated with EBRT and HDR-ICR have not been established.
For the current study, we retrospectively evaluated dosimetric parameters that were correlated with LRC ≥ Grade 2 in patients with uterine cervical carcinoma who were treated with a combination of EBRT and HDR-ICR. The objective of this study was to determine appropriate dose estimates to the rectum that are predictive for LRC ≥ Grade 2.
MATERIALS AND METHODS
Between July, 1994 and September, 2002, 202 patients with previously untreated uterine cervical carcinoma were referred to our department for definitive RT. Among them, 157 patients who had complete dosimetric data available, who had clinical Stage IB–IIIB disease according to the International Federation of Gynecology and Obstetrics criteria, and who had a minimum follow-up of 6 months were included in this retrospective analysis. The study was performed in accordance with the guidelines of our Institutional Review Board, which determined that informed consent was not required.
Patient characteristics are shown in Table 1. The median age was 59 years (range, 29–81 yrs). The histologic disease types were squamous cell carcinoma in 146 patients (93%), adenocarcinoma in 7 patients (4.5%), and carcinomas not otherwise specified in 4 patients (2.5%). The staging work-ups included gynecologic examination, chest radiograph, intravenous pyelography, blood chemistries, cystoscopy, rectosigmoidoscopy, and abdominopelvic computed tomography (CT) scan. The stage distribution of patients was as follows: Twenty-five patients had Stage IB disease, 36 patients had Stage IIA disease, 83 patients had Stage IIB disease, 5 patients had Stage IIIA disease, and 8 patients had Stage IIIB disease. Eighty-six patients (54.8%) had tumors that measured < 5 cm in greatest dimension, and 71 patients (45.2%) had tumors that measured ≥ 5 cm in greatest dimension. No patients had a history of prior pelvic surgery.
Table 1. Patient Characteristics
No. of patients (%)
NOS: not otherwise specified.
Age in years
< 60 yrs
≥ 60 yrs
Squamous cell carcinoma
< 5 cm
≥ 5 cm
EBRT was delivered by linear accelerator with a 10-megavolt X-ray using the 4-field box technique in daily fractions of 1.8–2.0 grays (Gy), 5 days per week, up to a total dose of 41.4–66.0 Gy (median, 54.0 Gy). A 4-cm width midline shielding of 5-half value layer thickness was placed at anterior and posterior fields after a 36–50.4 Gy (median, 50.4 Gy) external dose. HDR-ICR was performed using a 60Co (Ralstron HDR Unit®; Shimatszu, Kyoto, Japan) or 192Ir (microSelectron HDR®; Nucletron, Veenendaal, the Netherlands) remotely controlled afterloading system. HDR-ICR commenced after midline shielding of EBRT, except under unsuitable conditions for intracavitary insertion (mainly due to poor tumor regression). In those patients, HDR-ICR was performed after the completion of EBRT. The total dose of HDR-ICR to Point A, which was defined as 2 cm cephalad and 2 cm lateral to the cervical orifice along the plane of the tandem, ranged from 21 Gy to 39 Gy (median, 28 Gy) and was delivered in 3.5–5.0 Gy fractions twice weekly. EBRT and HDR-ICR never were administered on the same day. The median treatment time for the total RT course was 69 days (range, 48–147 days). Twenty-one patients (13.4%) received concurrent chemotherapy (weekly cisplatin or cisplatin and 5-fluorouracil) as a part of the definitive treatment.
The applicator system consisted of an intrauterine tandem with 3 different angles (0 degrees, 15 degrees, and 30 degrees) and paired colpostats. Anterior and posterior vaginal packing was used to maximize the distance from the source to the anterior bladder and the posterior rectal wall. A Foley catheter was inserted and ballooned with contrast media (7 mL) to localize the bladder neck. Oral contrast solution (60 mL) was injected into the rectum to visualize the rectum and the distal sigmoid colon. Orthogonal radiographs were taken for dose calculations and identification of bladder reference points (BP) and rectal reference points (RP), as defined in International Commission on Radiation Units and Measurements Report 38 (ICRU 38) (Fig. 1). To calculate the radiation dose at the highest dose region of the rectum, 4–8 points (median, 6 points), including the RP, were marked in 1-cm steps along the anterior rectal wall (Fig. 1B). The highest dose point in the rectum among those points was considered the maximal point (MP).18 Of 157 patients, 119 patients (75.8%) received a higher dose at the MP than at the RP. The median values (mean ± standard deviation) of the RP ratio (the dose at the RP/the prescribed [Point A] ICR dose) and the MP ratio (the dose at the MP/the prescribed [Point A] ICR dose) were 63.0% (65.3% ± 23.4%) and 89.7% (89.8% ± 33.8%), respectively.
The total biologically effective dose (BED) at specific points, such as Point A (BEDPoint A), RP (BEDRP), and MP (BEDMP), was determined by a summation of the EBRT and HDR-ICR components. The following equation19 was used to calculate the total BED: BED = total dose × [1 + dose of fraction/(α/β)]; and total BED = BEDEBRT + BEDHDR-ICR,15, 16, 20 where an α/β value of 3 was used for late responding tissues. Based on this equation, BEDPoint A ranged from 111.7 Gy3 to 193.3 Gy3 (median, 149 Gy3), BEDRP ranged from 63.3 Gy3 to 206.2 Gy3 (median, 118 Gy3), and BEDMP ranged from 63.3 Gy3 to 275.6 Gy3 (median, 139.4 Gy3).
Evaluation of LRC and Follow-Up
Patient follow-up was performed every 3 months in the first 3 years after the completion of treatment and every 6 months thereafter. LRC was assessed by a review of radiation oncology charts, nursing notes, hospital discharge summaries, and referring physician notes. LRC arising from RT was classified according to the Radiation Therapy Oncology Group late radiation morbidity scoring criteria for the rectum (Table 2). Only LRC ≥ Grade 2 was included in the study, because accurate scoring of LRC < Grade 2 is difficult, and clinical intervention is not required. The median follow-up for all patients in this study was 29.1 months (range, 6.1–111.1 mos).
Table 2. Radiation Therapy Oncology Group Late Radiation Morbidity Scoring Criteria for the Rectum
Mild diarrhea; mild cramping; bowel movement 5 times daily; slight rectal discharge or bleeding
Moderate diarrhea and colic; bowel movement >5 times daily; excessive rectal mucus or intermittent bleeding
Obstruction or bleeding requiring surgery
The Fisher exact test and the Student t test were used to compare the distribution of clinical and dosimetric parameters between patients who developed LRC ≥ Grade 2 and patients who did not. Correlations between dosimetric parameters were assessed using the Pearson correlation coefficient test. Overall survival, pelvic control, and LRC rates were calculated using the Kaplan–Meier method. All time intervals were measured from the first day of treatment. The endpoint was LRC ≥ Grade 2 and was expressed, analyzed, and compared as an actuarial rate. If the patient did not develop LRC ≥ Grade 2 by the last follow-up, then the status was censored. Age (younger than 60 yrs vs.60 yrs or older), tumor size (< 5 cm vs. ≥ 5 cm), disease stage (Stage IB–IIA vs. Stage IIB–IIIB), concurrent chemotherapy (no vs. yes), ICR fraction size (< 4 Gy vs. ≥ 4 Gy), RP ratio (< 80% vs. ≥ 80%), MP ratio (< 90% vs. ≥ 90%), EBRT dose (< 60 Gy vs. ≥ 60 Gy), BEDPoint A (< 160 Gy3 vs. ≥ 160 Gy3), BEDRP (< 125 Gy3 vs. ≥ 125 Gy3), BEDMP (< 160 Gy3 vs. ≥ 160 Gy3), and LRC severity (≤ Grade 2 vs. ≥ Grade 2) were considered binary variables. The differences between those parameters were examined using a log-rank test. In addition, multivariate analysis was performed using Cox proportional hazards model to analyze correlations between the parameters described above and the probability of LRC ≥ Grade 2 using a stepwise, forward procedure. All statistical tests were 2-sided and were performed using SAS software programs (version 8.0.1; SAS Institute Inc., Cary, NC). A P value ≤ 0.05 was considered to indicate a statistically significant difference.
The 5-year actuarial overall survival and pelvic control rates for the entire population were 71.1% and 84.0%, respectively. Eighteen of 157 patients (11.5%) developed Grade 2 LRC, 1 patient (0.6%) developed Grade 3 LRC, and 1 patient (0.6%) developed Grade 4 LRC. There was no fatal Grade 5 LRC. The median interval from the start of treatment to the onset of LRC was 18.2 months (range, 5.0–43.8 mos). The 5-year actuarial overall rate of LRC ≥ Grade 2 in all patients was 18.4%.
The distributions of clinical and dosimetric parameters were compared between patients who did or did not develop LRC ≥ Grade 2 (Table 3). We found there was no significant difference between the 2 groups regarding the distribution of clinical parameters such as age (younger than 60 yrs vs. 60 yrs or older), tumor size (< 5 cm vs. ≥ 5 cm), disease stage (Stage IB–IIA vs. Stage IIB–IIIB), and concurrent chemotherapy (no vs. yes). In contrast, the following dosimetric parameters were higher in the group that developed LRC ≥ Grade 2 compared with the group that did not: RP ratio, EBRT dose, BEDPoint A, BEDRP, and BEDMP (P < 0.05). The ICR fraction size and the MP ratio did not differ between the two groups.
Table 3. Comparison of the Clinical and Dosimetric Parameters between Two Groups that Did and Did Not Develop Late Rectal Complication ≥ Grade 2
Late rectal complication
< Grade 2 (n = 137)
≥ Grade 2 (n = 20)
ICR: intracavitary radiotherapy; Gy: grays; RP ratio: (dose at rectal point/dose at Point A) × 100; MP ratio: (dose at maximal rectal point/dose at Point A) × 100; EBRT: external beam radiation therapy; BEDPoint A: biologically effective dose at Point A for late-responding normal tissues; BEDRP: BED at the International Commission on Radiation Units and Measurements reference point for the rectum; BEDMP: BED at maximum point for the rectum; SD: standard deviation; Gy3: a unit of biologically effective dose for late responding tissues as the α/β value is 3.
The four differing dosimetric parameters were intercorrelated with one another: EBRT dose versus BEDPoint A (correlation coefficient [r] = 0.580; P < 0.001); BEDPoint A versus BEDRP (r = 0.517; P < 0.001), BEDPoint A versus BEDMP (r = 0.509; P < 0.001), and BEDRP versus BEDMP (r = 0.749; P < 0.001). In addition, the RP ratio, MP ratio, rectumICRU, and rectumMP correlated with each other: BEDRP versus RP ratio (r = 0.576; P < 0.001), BEDMP versus MP ratio (r = 0.521; P < 0.001), and RP ratio versus MP ratio (r = 0.660; P < 0.001).
Univariate and multivariate analyses were performed to identify correlations between clinical and dosimetric parameters and the actuarial rate of LRC ≥ Grade 2 (Table 4). Univariate analysis showed that the RP ratio, MP ratio, EBRT dose, BEDPoint A, BEDRP, and BEDMP were correlated with LRC ≥ Grade 2. The 5-year actuarial rates of LRC ≥ Grade 2 were 8.7% in patients with an RP ratio < 80% and 36.6% in patients with an RP ratio ≥ 80% (P < 0.001), 10.9% in patients with an MP ratio < 90% and 34.6% in patients with an MP ratio ≥ 90% (P = 0.004), 13.3% in patients with an EBRT dose < 60 Gy and 35.9% in patients with an EBRT dose ≥ 60 Gy (P = 0.035), 13.7% in patients with a BEDPoint A < 160 Gy3 and 31.7% in patients with a BEDPoint A ≥ 160 Gy3 (P = 0.039), 5.4% in patients with a BEDRP < 125 Gy3 and 36.1% in patients with a BEDRP ≥ 125 Gy3 (P < 0.001) (Fig. 2), and 13.8% in patients with a BEDMP < 160 Gy3 and 26.4% in patients with a BEDMP ≥ 160 Gy3 (P = 0.012). Other parameters, such as age, tumor size, disease stage, concurrent chemotherapy, and ICR fraction size, were not associated with LRC ≥ Grade 2 (P > 0.05). Multivariate analysis showed that, of all clinical and dosimetric parameters, only BEDRP was correlated with LRC ≥ Grade 2 (P = 0.009) (Table 4).
Table 4. Analysis of Correlations between Clinical and Dosimetric Parameters and Actuarial Rates of Late Rectal Complication ≥ Grade 2
% Five-yr actuarial rate of late rectal complication ≥ Grade 2
ICR: intracavitary radiotherapy; Gy: grays; RP ratio: (dose at rectal point/dose at Point A) × 100; MP ratio: (dose at maximal rectal point/dose at Point A) × 100; EBRT: external beam radiation therapy; BEDPoint A: biologically effective dose at Point A for late-responding normal tissues; BEDRP: BED at the International Commission on Radiation Units and Measurements reference point for the rectum; BEDMP: BED at maximum point for the rectum; Gy3: a unit of biologically effective dose for late responding tissues as the α/β value is 3.
Multivariate analysis using the Cox proportional hazard model.
< 60 yrs
≥ 60 yrs
< 5 cm
≥ 5 cm
Fraction size of ICR
< 4 Gy
≥ 4 Gy
< 60 Gy
≥ 60 Gy
< 160 Gy3
≥ 160 Gy3
< 125 Gy3
≥ 125 Gy3
< 160 Gy3
≥ 160 Gy3
LRC is one of the most common dose-limiting toxicities in patients who receive a combination of EBRT and HDR-ICR for uterine cervical carcinoma. Published studies have shown that the incidence of LRC varies in patients with uterine cervical carcinoma who are treated with a combination of EBRT and HDR-ICR. In a review of 24 HDR-ICR studies, Petereit and Pearcey21 noted that the median overall complication rate was 5%, which was much lower than that found in the current study and in other investigations.3, 15, 22–26 That low incidence probably was because the vast majority of those studies (22 of 24 studies) used a crude method of complication reporting that may have underestimated the true complication incidence. Other investigators3, 6, 15, 22–26 reported that the overall actuarial rate of LRC was 7–38%, comparable to what we found in the current study. Because LRC can result in bleeding that may require transfusion or hospitalization, bowel obstruction or perforation, fistula, and death, it would be very useful to be able to predict the likelihood of LRC accurately after EBRT and HDR-ICR in patients with uterine cervical carcinoma.
Some investigators5–8, 21, 22, 27 previously evaluated the effects of EBRT dose and total cumulative BEDPoint A on the risk of LRC. Hamberger and colleagues5 demonstrated that severe complication rates were 3.1%, 10%, and 15% in patients who received 40 Gy, 50 Gy, and 60 Gy whole-pelvic EBRT, respectively. Roeske et al.7 suggested that an EBRT dose ≥ 60 Gy and a BEDPoint A ≥ 135 Gy3 were predictors of LRC. Petereit and Pearcey21 noted that BEDPoint A was correlated with LRC rates for all stages combined but did not find the threshold dose for LRC. Because most centers treat patients who have uterine cervical carcinoma with the Point A dose, many previous studies5–8 did not analyze the influence of the BEDPoint A on the risk of LRC. In addition, some investigators9, 28 noted that the EBRT dose and BEDPoint A did not correlate with LRC. In the current study, although the EBRT dose and BEDPoint A were correlated with LRC ≥ Grade 2 in the univariate analysis, those parameters were not correlated in the multivariate analysis.
The American Brachytherapy Society recommends that attempts should be made to keep the RP ratio (dose at RP/prescribed ICR dose at Point A) below 80%, emphasizing the importance of the rectal dose given by ICR.20 Shin et al.29 noted that an RP ratio ≥ 80% and a total ICR dose ≥ 20 Gy correlated with a risk of LRC. Cheng and colleagues30 demonstrated that patients with a ratio > 90% between the proximal rectal dose and the prescribed ICR dose at the Point A (the MP ratio) were more likely to develop LRC. However, other investigators3, 10 reported that ICR parameters, such as total ICR dose, RP ratio, and dose rate, did not correlate with the risk of LRC. In the current study, although the ICR fraction size, RP ratio, and MP ratio were associated significantly with the risk of LRC ≥ Grade 2 in univariate analysis, those parameters were not associated significantly in the multivariate analysis.
Several investigators13–17, 31 have sought to establish a correlation between LRC and the total BEDRP. Conceptually, both EBRT and HDR-ICR can result in LRC in patients with cervical carcinoma; thus, the cumulative rectal dose potentially is more valuable for predicting the risk of LRC than other dosimetric parameters (EBRT dose, RP ratio, MP ratio, fraction size of ICR). In a study of 1211 patients with uterine cervical carcinoma who were treated with EBRT and LDR-ICR, Perez and coworkers13 demonstrated a correlation between a rectal dose > 80 Gy and LRC ≥ Grade 3. This dose corresponded to a BED of 120–128 Gy3. Clark and colleagues14 reviewed 43 patients with uterine cervical carcinoma who were treated with EBRT and HDR-ICR. They demonstrated a strong correlation between the risk of LRC ≥ Grade 3 and a BEDRP > 125 Gy3. Ogino and coworkers16 also reported a positive correlation between the risk of LRC (all grades) and BEDRP; LRC was observed in 0% of patients who received a BEDRP < 100 Gy3, in 14% of patients who received 100–120 Gy3, in 44% of patients who received 120–140 Gy3, and in 72% of patients who received 140 Gy3. Contrary to these findings, Ferrigno et al.22 and Kapp et al.3 reported that BEDRP was not correlated with LRC. Those investigators explained that the lack of correlation probably was because their median BEDRP was lower than that reported in other studies.13–17 In the current study, we observed a significant correlation between BEDRP and LRC ≥ Grade 2, with a 5-year actuarial rate of LRC ≥ Grade 2 of 5.4% for patients who received a BEDRP < 125 Gy3 and 36.1% for patients who received a BEDRP ≥ 125 Gy3 (P < 0.001).
Several investigators10–12 have suggested that the RP, as defined by ICRU 38, does not always represent the exact location of the greatest dose to the rectum. Pourquier and colleagues11 reported that doses to both the MP and the RP served as primary indicators for predicting the risk of LRC. Cheng and colleagues30 noted that the proximal rectal dose was greater than the RP dose and was correlated more significantly with the risk of LRC ≥ Grade 2 than the RP dose. The determination of the maximal RP based on the two-dimensional (2D) orthogonal method is difficult and may be inconsistent compared with the ICRU RP because of the inadequate visualization of the whole rectum up to the rectosigmoid junction in orthogonal radiographs and the varying anatomy of the proximal rectum and sigmoid region. Compared with the more recent 3D planning techniques, 2D planning generally overestimates the minimum dose delivered to the target volume and underestimates the maximum dose to the normal tissues.12, 32–34 Consequently, reliable estimates of doses to the rectum for predicting the risk of LRC have not been established to date, and further larger scale studies and careful analyses of dose-volume histograms are required. However, the effectiveness of the absorbed dose at the RP, as defined by ICRU 38, has been demonstrated in other reports, with correlations between the dose estimates at the RP (BEDRP) and the risk of LRC being identified.10–12, 31 In the current study, univariate analysis showed that all dosimetric parameters (RP ratio, MP ratio, EBRT dose, BEDPoint A, BEDRP, and BEDMP) were associated with the risk of LRC ≥ Grade 2 and that all were interrelated closely. Although further studies will be needed to establish which dosimetric parameter is the best, the most reproducible, or representative to predict the risk of LRC ≥ Grade 2, our results showed that BEDRP was the most significant among these dosimetric parameters in multivariate analysis.
In conclusion, for the current study, we examined the correlation between dosimetric parameters and LRC ≥ Grade 2 in patients with uterine cervical carcinoma who received a combination of EBRT and HDR-ICR. Our findings indicate that BEDRP is associated significantly with the risk of LRC ≥ Grade 2. Coverage of the target volume should not be compromised by fear of LRC, because pelvic recurrence is always the worst complication in patients with uterine cervical carcinoma who are treated with a combination of EBRT and HDR-ICR. However, based on our current results, we suggest that maximal effort should be made to keep BEDRP below 125 Gy3 whenever possible to minimize the risk of LRC ≥ Grade 2 in patients with uterine cervical carcinoma who are treated with EBRT and HDR-ICR.
The authors thank Mrs. Young-Joo Shin and Mr. Rho-Bong Myung (Department of Radiation Oncology at Gil Medical Center) for help with collecting patient data.