The authors previously reported that the mitotic index of a proliferating cell population (pMI) was a potent prognostic factor in cervical cancer patients treated with photon beam therapy. In this study, they investigated whether the pMI accurately predicted prognosis in cervical cancer patients treated with carbon ion beam.
Tissue sections were obtained from 27 consecutively treated patients with stage IIIB bulky (19 patients) and stage IVA (8 patients) squamous cell carcinomas of the cervix treated with carbon ion beam at the National Institute of Radiological Sciences, Japan, as a phase I/II study with dose escalation methodology (52.8-72 grays equivalent radiation dose/24 fractions). The mitotic index (MI) and Ki-67 labeling index (Ki-67-LI) were determined by hematoxylin and eosin staining and immunohistochemical staining, respectively. The pMI was calculated using the following formula: pMI = MI/Ki-67-LI.
The pMI ranged from 0.6 to 8.9 (mean, 3.9 ± 2.6; median, 3.2). Twelve of the 27 specimens had a pMI >3.5. The local control rate in tumors with a pMI >3.5 was 17%, significantly lower than the 73% in the tumors with a pMI <3.5 (P = .005). Multivariate analysis indicated that the pMI had the strongest impact on local control (standard regression coefficient = 0.48, P = .002) among the variables, including clinical stage, irradiated dose, age, and tumor volume.
Heavy charged particle radiation therapy for cancer treatment began at the National Institute of Radiological Sciences (NIRS, Chiba, Japan) in June 1994 using carbon ions generated by the heavy-ion medical accelerator in Chiba, and more than 3000 patients have been treated to date.1 Several reports have demonstrated the favorable results of carbon ion beam radiotherapy (CIRT) in the treatment of malignant tumors, including head and neck cancer,2 stage I nonsmall cell lung cancer,3, 4 hepatocellular carcinoma,5 choroidal melanoma,6 prostate cancer,7, 8 bone and soft tissue sarcomas,9 and sacral chordomas.10
In terms of cervical cancer, a phase I/II clinical trial for advanced cancers of the uterine cervix (protocol 9403) began in June 1995 and ended in February 1998. The preliminary results of those clinical trials have been reported by Nakano et al11 and Kato et al.12 These reports have shown that carbon ion beam therapy for cervical cancer is feasible,11 and the local control rate of patients treated with ≥62.4 grays equivalent radiation dose (GyE) was favorable even for the patients with stage IVA disease, or for those with tumors ≥6.0 cm.12 Furthermore, we have also reported similar disease-free survival and local control rates between hypoxic and oxygenated tumors before and during treatment, indicating that the tumor oxygenation status is relatively unimportant in local control in carbon beam therapy.13
The proliferative activity of tumor cells has been determined immunohistochemically by Ki-67 antigen, as well as other markers. The Ki-67 labeling index (Ki-67-LI) has been reported to be a significant prognostic factor in many malignant tumors.14-22 The mitotic index (MI; percentage of mitotic cell) of tumor cells also correlates to the cell proliferation speed, and it too has been reported as a significant prognostic factor.21-29 We have reported that Ki-67-LI and MI correlate with prognosis in patients with cervical cancer treated with photon radiation therapy.30 However, the mitotic index in vivo is sometimes biased by the presence of quiescent cells whose population is larger than the cycling cell population in human cancers.30 We have proposed that the MI of a proliferating cell population (pMI) can express the relative cell cycle speed and can be estimated by the counting of the MI and Ki-67-LI.30 Furthermore, we have reported that the pMI correlates well with the prognosis of patients with cervical cancer who received photon beam therapy.30-32 The pMI was found to be a significantly stronger prognostic factor than Ki-67-LI.31
Carbon ion beam therapy has various biologic advantages in terms of high linear energy transfer (LET) radiation. Potentially lethal and sublethal damage repair is attenuated by high LET radiation.33 Blakely et al reported that high LET beam diminished cell cycle–dependent radiosensitivity compared with those observed with low LET radiation.34 Barendsen et al reported that high LET radiation is apparently most effective on tumors with a large G0 fraction.35 As for tumor oxygen status, which is an important issue in radiation therapy, high LET radiation induces cell death independently of the tissue oxygen status, as we previously demonstrated.14 Furthermore, Takahashi et al36 reported that the carbon ion beam can induce apoptosis regardless of p53 status. Hence, the prognostic factor for high LET radiation therapy may be completely different from that of low LET radiotherapy. In this study, we investigated whether Ki-67-LI, MI and/or pMI predicted prognosis in cervical cancer patients treated with a carbon ion beam.
MATERIALS AND METHODS
Patient Characteristics and Specimens
Twenty-seven patients with stage IIIB bulky and stage IVA squamous cell carcinomas of the cervix were treated with carbon ion beam at the NIRS between 1995 and 1998 as a phase I/II study in a dose escalation manner (protocol: 9403). Their ages ranged from 36 to 72 years (mean ± standard deviation [SD], 56.2 ± 9.3; median, 54 years). All the tumors had a maximum diameter >50 mm. Tumor size was assessed by both pelvic examination and magnetic resonance imaging (MRI), and dimensions of the cervical tumor were measured according to T2-weighted MRI images.37, 38 The mean ± SD and median tumor volume were 153 ± 113 and 109 mm3, respectively. Of these, 19 patients had stage IIIB disease, and 8 had stage IVA disease, all of which involved bladder invasion, except 1 case with rectal invasion. The clinical staging and histologic classification were based on the criteria of the International Federation of Gynecology and Obstetrics and the World Health Organization.39, 40 All patients were followed for a minimum of 5 years or until death. Tissue sections were obtained from all 27 patients before treatment. They were excised from the cervical tumors and fixed in 10% formaldehyde for approximately 24 hours and embedded in paraffin.
The disease status and treatment methods were explained carefully and precisely to the patients by the treating physicians. In addition, an explanation of the tumor biopsy measurement procedure was provided to all patients who qualified for the study. After these explanations, written informed consent was obtained from the patients and the family, according to the institutional regulations.
CIRT was performed at the National Institute of Radiological Sciences in Chiba, Japan. Details of the treatment protocol have been previously reported by Kato et al12 and Nakano et al.13 Briefly, the total number of treatment fractions and the time period were fixed at 24 fractions over 6 weeks with 4 fractions per week. Anteroposterior and posteroanterior ports were used for 16 fractions over 4 weeks to irradiate cervical tumors and pelvic lymph node chains. An additional 8 fractions over 2 weeks were given by lateral opposing ports to boost only cervical tumors. As this represents a phase I/II study with a dose escalation methodology, the treatment was initiated with a fraction dose of 2.2 GyE,41 and the fraction doses increased by 0.2 GyE per step to 2.4 GyE, 2.6 GyE, 2.8 GyE, and 3.0 GyE. Therefore, the initial dose of 52.8 GyE was increased by 4.8 GyE per step to a total of 72.0 GyE. Each dosing group consisted of at least 5 patients. The energy of the carbon ion beam ranged from 350 MeV to 400 MeV.
Measurements of pMI, MI, and Ki-67-LI
The tissue sections, placed on silane-coated microslides, were deparaffinized and dehydrated.
For measurement of the Ki-67-LI, specimens were immunohistochemically stained (Fig. 1). Specimens were heated to >95°C for 20 minutes in 0.01 M citrate buffer, pH 6, using a Microwave Processor (H2800, Energy Beam Sciences Inc., Agawam, Mass) to unmask the antigens. After the unmasking, the sections were cooled at room temperature for 1 hour. Endogenous peroxidases were blocked with 3% hydrogen peroxide for 10 minutes, and the sections were washed 3 times with phosphate-buffered saline (PBS). They were then incubated overnight at 4°C with anti-Ki-67 (MIB-1, Immunotech International, Marseilles, France). After incubation, they were washed with PBS 3 times. They were incubated with a labeled-polymer–conjugated second antibody, an Envision kit (DAKO, Carpinteria, Calif), for 30 minutes. They were washed with PBS and then developed with 3,3′-diaminobenzidine tetrahydrochloride for 5 minutes at room temperature and then lightly counterstained with hematoxylin, dehydrated, and mounted. As a negative control, a specimen was treated without a primary antibody in the incubation step. The Ki-67-LI was calculated as the percentage of Ki-67–positive cancer cells by counting >500 cancer cells. A cutoff value for the Ki-67-LI was defined as 33%.
For measurement of the MI, specimens were stained with hematoxylin and eosin. The MI was calculated as the percentage of mitotic cancer cells by counting >2000 cancer cells. A cutoff value for the MI was defined as 1.5%.
The pMI was calculated with the following formula:
M indicates mitotic cell population (the number of mitotic cells); P, proliferating cell population (the number of Ki-67–positive cells); Q, quiescent cell population (the number of all cells − the number of Ki-67–positive cells).
A cutoff value for the pMI was defined as 3.5, based on the results of our previous report.30 A cutoff value for the MI was established as a positive finding.
The Kaplan-Meier method was used for the local control and metastasis-free survival rates, and differences were statistically analyzed with the log-rank test. Mean values of pMI, MI, and Ki-67-LI by irradiated dose group were compared by 1-factor analysis of variance. Univariate analysis for local control and metastasis-free survival was applied with Pearson correlation coefficient. Multivariate analysis for local control and metastasis-free survival was applied with the Cox proportional hazard model with a 95.0% confidence interval. The differences were considered statistically significant at P < .05. All analyses were performed with StatView (Version 5.0, SAS Institute Inc, Cary, NC).
The pMI ranged from 0.6 to 8.9 (mean ± SD, 3.9 ± 2.6; median, 3.2). A total of 44% (12 of 27) of the tissue specimens had a pMI >3.5. The MI ranged from 0.14% to 3.3% (mean ± SD, 1.23 ± 0.83%; median, 1.02%). A total of 33% (9 of 27) of the tissue specimens had a >1.5% MI. The Ki-67-LI ranged from 15% to 80% (mean ± SD, 35 ± 15%; median, 29%). A total of 44% (12 of 27) of the tissue specimens had a Ki-67-LI >33%. Table 1 shows the means value of pMI, MI, and Ki-67-LI by dose group. There was no significant difference among the dose group. Although there was no significant difference, pMI and MI in the low dose group (≤62 GyE) were bigger than in the high dose group (≥67 GyE).
Table 1. Mean Values of pMI, MI, and Ki-67-LI by Dose Group
pMI indicates mitotic index of proliferating cell populations; MI, mitotic index; Ki-67-LI, Ki-67 labeling index; GyE, gray equivalent radiation dose.
Nine of 12 patients with a >3.5 pMI had local recurrence, compared with only 4 of 15 patients with a pMI <3.5. The local control rate of the tumors with a pMI >3.5 was 17%, significantly lower than the 73% of the tumors with a pMI <3.5(P = .005; Fig. 2). The local control rate of the tumors with a >1.5% MI was 17%, significantly lower than the 66% of the tumors with a <1.5% pMI (P = .02). The local control rate of the tumors with a Ki-67-LI >33% was 64%, which was higher than the 37% of the tumors with a Ki-67-LI <33%, although the difference was not significant (P = .13).
The metastasis-free survival rate of the tumors with a >3.5 pMI was 23%, lower than the 58% of the tumors with a <3.5 pMI, although the difference was not significant (P = .26; Fig. 3). The metastasis-free survival rate of the tumors with an MI >1.5% was 23%, lower than the 54% of the tumors with an MI <1.5%. The difference was not significant (P = .60). The metastasis-free survival rate of the tumors with a >33% Ki-67-LI was 63%, higher than the 34% of the tumors with a <33% Ki-67-LI, although the difference was not significant (P = .14).
Table 2 shows the univariate analysis for local control and metastasis-free survival. The pMI and MI significantly correlate to the local control (correlation coefficient, 0.481 and 0.419, respectively). Among Ki-67-LI, MI, and pMI, pMI was the strongest factor for local control. None of these factors significantly correlates to metastasis-free survival. Table 3 shows the multivariate analysis for local control and metastasis-free survival. The pMI had robustly significant impact on local control (standard regression coefficient = 0.484, P = .021), but no significant impact on metastasis-free survival (standard regression coefficient = 0.131, P = .54) among the variables, including irradiated dose (52.8 GyE vs 57.6 GyE vs 62.4 GyE vs 67.2 GyE vs 72.0 GyE), clinical stage (stage IIIB vs stage IVA), age (≤50 years old vs >50 years old), and tumor volume (≤100 cm3 vs >100 cm3).
Table 2. Correlation Matrix for Local Control and Metastasis-free Survival
Local Control (P)
Metastasis-free Survival (P)
pMI indicates mitotic index of proliferating cell populations.
Ki-67 labeling index
Table 3. Multivariate Analysis for Local Control and Metastasis-free Survival
Standard Regression Coefficient
GyE indicates gray equivalent radiation dose; pMI, mitotic index of proliferating cell populations.
Local control (recurrence/control)
Irradiation dose (52.8/57.6/62.4/67.2/72.0 GyE)
Clinical stage (IIIB/IVA)
Age (≤50 y)
Tumor volume (≤100 cm3)
Metastasis-free survival, meta (+)/meta (−)
Irradiation dose (52.8/57.6/62.4/67.2/72.0 GyE)
Clinical stage (IIIB/IVA)
Age (≤50 y)
Tumor volume (≤100 cm3)
The MI and Ki-67-LI are well-known cell growth-associated parameters and have been reported to be useful and powerful prognostic factors.14-29 In general, a high MI and high Ki-67-LI correlate with a worse prognosis. However, in the course of radiation therapy, tumors with higher proliferative activity tend to have higher radiosensitivity and better prognosis.42 Furthermore, as mentioned earlier, the mitotic index in vivo is sometimes biased by the presence of quiescent cells, the population of which is larger than the cycling cell population in human cancers. Therefore, we have proposed the pMI as a more accurate cell growth–associated parameter and reported that pMI strongly correlates with prognosis in patients with cervical cancer who have received photon beam therapy.30-32
In this study, the local control rate of the tumors with a pMI >3.5 was only 17%, although most of the tumors were bulky. To improve the local control rate of radiation therapy, 1) shortening the treatment period (accelerated fractionation),43, 44 and 2) combined use of chemotherapy45, 46 have been investigated. We have reported a phase I/II clinical study of carbon ion radiotherapy for locally advanced carcinoma of the uterine cervix.12, 13 We have fixed the fraction number at 24 fractions in a 6-week period, and an accelerated-fractionated protocol has not yet been planned, because of the relatively high probability of late adverse effects. However, heavy ion beam therapy has the advantage of better dose distribution and biological effect, and the excellent dose distribution of the heavy ion beam enables a reduction in the irradiation volume absorbed by normal tissue. Also, biologically, the therapeutic ratio increases if short-course accelerated-fraction schemes are used in carbon ion beam therapy.1 Hence, accelerated-fractionated radiotherapy is desirable. In addition, although the “redistribution,” “recruitment,” and “reoxygenation” of the tumor during radiation therapy are benefits of a fractionated treatment schedule of conventional photon beam therapy, the benefits of fractionation of the high LET beam are believed to be less than the low LET beam, because high LET beam diminishes both the cell cycle–dependent radiosensitivity and oxygen enhancement ratio compared with low LET radiation.34, 47 At the NIRS in Japan, accelerated-fractionated carbon ion beam therapy has been investigated systematically for a variety of tumor entities, and it has been indicated that a significant reduction of overall treatment time can be accomplished for many tumor entities without enhancing toxicity.1 Miyamoto et al4 reported that the treatment regimen of accelerated-fractionated CIRT for stage I lung cancer was 72 GyE/9 fractions in 3 weeks, in which the local control rate was >90%, and the rate of late adverse effects of grade 2 or more severe was only 8%. Furthermore, a dose escalation study on the single-fraction treatment has been initiated.1 Although cervical cancer has been treated with 24 fractions over 6 weeks and an accelerated-fractionated protocol has not yet been planned because of the relatively high probability of late adverse effects, prostate, which is in close proximity to the rectum and bladder, as in the case of the uterus, has been treated with 66 GyE/20 fractions over 5 weeks, and the rate of late adverse effects was only 7%, and the local control rate was 100%.7 Hence, although some improvement of treatment is still needed, accelerated-fractionated CIRT for cervical cancer appears feasible.
In the present study, no significant correlations were detected between pMI, MI, and Ki-67-LI and metastasis-free survival rate. However, there was a large difference of approximately 35% in the metastasis-free survival rate between the patients with the low and high pMI tumors. As shown in our recent report,12 although the local tumor control was relatively good, distant metastases did frequently occur, and the 5-year overall survival rate was still unsatisfactory. To improve the survival rate as well as the local control rate, the use of chemotherapy in combination with CIRT should be further explored, especially for patients with high pMI tumors, although further study is still required to properly assess the significance of pMI for metastasis-free survival with higher patient numbers.
In conclusion, we investigated whether Ki-67-LI, MI, and/or pMI predicted prognosis in cervical cancer patients treated with a carbon ion beam. The results of this study suggest that a high pMI is indicative of a poor prognosis in patients with squamous cell carcinomas of the cervix treated with carbon ion beam therapy. Although further studies with larger data sets is warranted, to improve the local control rate of CIRT for cervical squamous cell carcinoma, especially for those tumors having a high pMI, an accelerated-fractionated regimen together with the use of chemotherapy should be further explored as explored in photon beam therapy.43-46