The Relationship Between the Radiation Dose of Different Anatomic Bony Sites and Neutrophil Toxicity in Concurrent Chemoradiotherapy for Cervical Cancer

Background: The purpose of this study was to verify the radiation dose of the pelvic bone marrow of different anatomical bony sitesand the incidence of neutrophil toxicity during the concurrent chemoradiotherapy for cervical cancer. Methods: There were 117 cervical cancer patients who received concurrent chemoradiotherapy included in this research. The radiotherapy includedexternal-beam radiation therapy (EBRT) and the brachytherapy. The dosimetric parameters included V5, V10, V20, V30, V40, V50, and Dmean. The nal neutrophil count was dened as the lowest neutrophil count after 2 circles of concurrent chemotherapy, during or within 1 month after the treatment. The correlation between the dosimetric parameters and the lowest neutrophil count were analyzed by linear regression, the cutoff values of the dosimetric parameters were obtained using the ROC curves,the patients were divided into subgroups based on the cutoff values. The clinicopathologic features and the dosimetric parameters were included into the multivariate regression analysis model to further prove the correlation between the dosimetric parameters and the neutrophil toxicity. Data were assessed with IBM-SPSS software version 22.0, and all values of p<0.05were considered statistically signicant. Results:The neutrophil toxicity (grade 1-4) rate was 58.97%. The linear regression showed the Dmean and V50 of Lumbosacral vertebrae(LS), the Dmean, V5, V10, V20, V30, V40 and V50 of the ilium correlated with the lowest neutrophil count, while none of the dosimetric parameters of the femoral correlated with the lowest neutrophil count.The multivariate analysis showed the V20, V30 and V50 of the LS, the Dmean, V5, V10, V20 and V30 of the ilium correlated well with the neutrophil toxicity; none of the dosimetric parameters of the femoral correlated with the neutrophil toxicity. Conclusion:During cervical cancer, of

chemoradiotherapy, hematologic toxicity is still a common clinical problem, while the neutrophil toxicity is the most common in all of the hematologic adverse events. When the neutrophil toxicity happens, it increases the risk of fever and infection, and sometimes, it might affect the intensity and progress of radiotherapy and chemotherapy, which may ultimately lead to poor outcome [7][8][9] . There have been many studies that proved the risk of hematologic adverse events correlated with the high dose and volume of pelvic bone marrow, and almost authors suggested to limit the dose and volume of bone marrow irradiation to reduce the risk of hematologic adverse events [10,11] . Several investigators have begun to consider the amount and distribution of low-dose radiation to the bone marrow as a part of their IMRT optimization process, so-called bone marrow-sparing IMRT. [12][13][14][15][16] However, the hematopoietic function of bones depends on the content of their active bone marrow. Researchers have already proved the proportion of proliferating bone marrow is different in different anatomical sites of bones. [17][18][19] The radiation dose of different anatomical skeletal sites might in uence the hematologic toxicity differently. Therefore, we retrospectively analyzed 117 cervical cancer patients who received concurrent chemoradiotherapy, to prove the radiation dose of different anatomical sites in the pelvic radiotherapy in uence the neutrophil toxicity in varying degree.
Material And Methods

Patients
We retrospectively analyzed the patients who received concurrent chemoradiotherapy between January 2016 to December 2018 in Tianjin Medical University Cancer Hospital, all of the patients were pathologically diagnosed cervical squamous carcinoma or adenocarcinoma, The inclusion criteria were: (1) staged IB-IIIC (based on Federation International of Gynecology and Obstetrics staging, Figo); (2) aged 18-70 years; (3) Karnofsky performance status score ≥ 70; (4) complete records of blood routine examination at the time of pretreatment, weekly during the treatment and within a month after the treatment. The exclusion criteria were : (1) the concurrent chemotherapy nished less than 3 cycles; (2) the interrupt interval of radiotherapy longer than one week for any reason; (3) second primary tumor; (4) history of radiotherapy. There were 117 patients included into the study.

Radiotherapy
All of the patients received external-beam radiation therapy (EBRT) and the brachytherapy.
(1) The EBRT: Radiotherapy was performed using the linear accelerator, 6MV-X rays. Simulate patients prone with CT planning. The clinical target volume (CTV) includes gross tumor, entire uterine, cervix, vagina, the internal and external iliac lymph node drainage area, and the obturator lymph node drainage area, with a superior border at the lumbar 4/5 level, and inferior border at 3 cm below the most inferior vaginal involvement; The gross tumor target volume of lymph node (GTVnd) includes all of the de nitely diagnosed metastatic lymph nodes in pelvic cavity. The planning target volume of CTV (PTV) is to add 0.7 cm laterally and 1.5 cm axially on primary CTV, the planning target volume of GTVnd (PGTVnd) is to add 0.7 cm laterally and 1.5 cm axially on primary GTVnd. The total dose delivered to PGTVnd was 59.92 Gy(2.14 Gy per fraction, 28 fractions), the total dose delivered to PTV was 50.4 Gy(1.8 Gy per fraction, 28 fractions), and the radiation therapy progressed as 1 fraction/day*5days/week. The contouring of organ at risk include small bowel, rectum, bladder and the femoral heads.

The Concurrent Chemotherapy
The concurrent chemotherapy regimen was cisplatin(25 mg/m 2 ) weekly, started from the rst week of the EBRT, totally 5 circles as planned. The chemotherapy was stopped when the WBC count was lower than 2.0*109/L, or the neutrophil count was lower than 1.5*109/L.

Dosimetric Parameters Of The Bone Marrow
The contouring of the bone marrow(BM) included the lower lumbosacral vertebrae (the superior border depending on the superior border of the PTV), the ilium and the upper femur(the inferior border depending on the inferior border of the PTV). The volume dosimetric parameters included V5(Volume receiving 5 Gy and so on), V10, V20, V30, V40, V50, and Dmean(the mean dose of the BM). The effect of brachytherapy to BM was ignored.

The Statistical Analysis
The correlation between the dosimetric parameters and the lowest neutrophil count were analyzed by linear regression, the cutoff values of the dosimetric parameters were obtained using the ROC curves. The patients were divided into subgroups based on the cutoff values of the dosimetric parameters. The clinicopathologic features and the dosimetric parameters were included into the multivariate regression analysis model to further prove the correlation between the dosimetric parameters and the neutrophil toxicity. Data were assessed with IBM-SPSS software version 22.0, and all values of p < 0.05 were considered statistically signi cant.

The clinicopathologic features
There were 117 patients included into the study, with a median age of 54 (29-70) years, there were 83 cases with the age < 60 years and 34 cases with the age ≥ 60 years. Based on the pathology, there were 97 cases of Squamous carcinoma and 20 cases of adenocarcinoma. Based on the Figo stage, there were 9 cases of stage I, 79 cases of stage II and 29 cases of stage III. Based on the prescription radiation dose, there were 97 cases with the dose < 59.92 Gy and 20 cases with the dose ≥ 59.92 Gy. Data was showed in Table 1. The total neutrophil toxicity (grade 1-4) rate was 58.97%; The age, pathology, stage and prescription dose were taken into the multivariate analysis. .74 ± 28.43 ml respectively; the mean V40 of the three regions were 252.76 ± 76.14 ml, 227.87 ± 80.81 ml and 6.10 ± 6.63 ml respectively; the mean V50 of the three regions were 141.09 ± 49.97 ml, 97.11 ± 53.54 ml and 0.29 ± 0.85 ml respectively. The mean dose of the LS was higher than the other two regions, the volume of low and medium dose of the ilium was larger than the other two regions, the volume of high dose of the LS was higher than the other two regions. In the region of femoral, both of the mean dose and the volume of the low, medium and high dose were lower/smaller than the other two regions, while the V40 and V50 of the femoral were negligible compared with the other two regions. (Fig. 1-7) 3. The correlation between the dosimetric parameters and the lowest neutrophil count.
The dosimetric parameters including Dmean, V5, V10, V20, V30, V40 and V50 of the three anatomical region namely lumbosacral vertebrae(LS), ilium and femoral were taken into the linear regression. The Dmean and V50 of LS, the Dmean, V5, V10, V20, V30, V40 and V50 of the ilium were found to correlate with the lowest neutrophil count, while none of the dosimetric parameters of the femoral correlated with the lowest neutrophil count. The results were showed in Table 2. The patients were divided into subgroups based on the ROC lines for the the dosimetric parameters including Dmean, V5, V10, V20, V30, V40 and V50 of the LS, ilium and femoral respectively. Then all of the dosimetric parameters of the three regions were taken into the multivariate analysis for grade 1-4 neutrophil toxicity together with the age, pathology, stage, and prescription dose respectively. The V20, V30 and V50 of the LS correlated well with the neutrophil toxicity; the Dmean, V5, V10, V20 and V30 of the ilium correlated well with the neutrophil toxicity; none of the dosimetric parameters of the femoral correlated with the neutrophil toxicity. And the Figo stage(II) and the prescription dose were also correlated well with the neutrophil toxicity. The results were showed in Table 3 to 9.

Discussion
As the development of the radiotherapy, the conventional radiotherapy technique has been gradually replaced by more advanced techniques such as IMRT, Vmat, and so on, which effectively reduced the volume and dose of irradiation of the normal tissue, but low dose irradiation volume did not decrease [5,6] , there are still many issues of concern. The concurrent chemoradiotherapy has become the standard treatment for locally advanced cervical cancer [4] . But meanwhile the hematological toxicity is still a common adverse event in the radiotherapy of pelvic tumor, when it is serious, it will limit the intensity of radiotherapy and chemotherapy, prolong the duration of the treatment, and ultimately lead to adverse impact to the treatment [7][8][9][20][21][22] . It has been found that up to 50% of a patient's total active bone marrow is within the pelvis and lumbar spine [23] . So it is important to protect the bone marrow of the pelvis and lumbar spine during the radiation of the pelvic tumor. The relationship between bone marrow exposure and hematological toxicity during the pelvic radiation has been proved, and reducing the radiation dose of bone marrow could reduce the risk of clinical hematological toxicity has although been proved [24,25] . The published results found that BM sparing IMRT could reduce acute hematological toxicity for patients with locally advance cervical cancer, the dosimetric parameters used were the mean dose, V10 and V20 [26,27] . But do every dosimetric parameter of different anatomical bone regions in uence the hematological toxicity equally? Of course not. The hematopoietic function of bones depends on the content of their active bone marrow. Researchers have already proved the proportion of proliferating bone marrow varied in different anatomical sites of bones. [17][18][19] Even in the pelvic skeleton, which includes the inferior lumbar vertebrae, the sacral vertebrae, the ilium and the upper femur, the radiation dose of these different anatomical regions in uence the hematological/neutrophil toxicity differently. So in the process of setting the radiotherapy, the different anatomical skeletal regions should be treated differently as organs at risk.
In this research, we aimed to prove there were different correlation between different anatomical regions and the neutrophil toxicity. As our result showed, the mean dose of the LS was the highest, and the mean dose of the femur was the lowest; the volume of the high dose was the largest in the LS region, the volume of the low and medium dose was the largest in the ilium region, while all of the volume of dose in the femur region was negligible in compare with the other two regions. The LS vertebrae was close to the PTV/PGTV, so the volume of the high dose was higher than the other two regions, but the LS vertebrae is cylindrical bone, so in the cross sections, the area of the LS vertebrae was small, the isodose curves of low and medium dose avoided the LS in most conditions; The ilium is at bone, it is around the outside of the pelvic cavity, where the isodose curves of low and medium dose mostly distributed, so the volume of the low and medium dose of the ilium was the highest.
The femur is almost out of the pelvic cavity, and is the farthest from the center of the target area of radiation, so all of the volume of dose in the femur region was negligible in compare with the other two regions. This might explained why the V20, V30 and V50 of the LS correlated well with the neutrophil toxicity, while the Dmean, V5, V10, V20 and V30 correlated well with the neutrophil toxicity, and none of the dosimetric parameters of the femoral correlated with the neutrophil toxicity. So we suggest that, in order to reduce the risk of neutrophil toxicity, the volume of medium and high dose of LS and the volume of low and medium dose of ilium should be strictly limited, the Dmean of the ilium should also be taken into consideration. While the dosimetric parameters of the femur could be ignored. However, this is a retrospective research, to get more precise results, the prospective randomized controlled study is needed.

Conclusion
During the process of concurrent chemoradiotherapy for cervical cancer, in order to reduce the risk of neutrophil toxicity, the volume of medium and high dose of LS and the volume of low and medium dose of ilium should be strictly limited, the Dmean of the ilium should also be taken into consideration. The dosimetric parameters of the femur could be ignored.