It has been reported that the incidence of grade ≥ 2 HT in uterine cervical cancer during chemoradiotherapy is 23-91.37% [5, 8, 21], and the incidence of grade ≥ 3 HT is 14.7-77% [10, 14, 21, 22]. In this study, the incidence of grade ≥ 2 leukopenia, neutropenia, thrombocytopenia and anemia for patients with uterine cervical/endometrial cancer during radiotherapy was 63.9%, 45.4%, 19.6%, and 38.8%, respectively. The incidence of grade ≥ 3 leukopenia, neutropenia, thrombocytopenia and anemia was 26.8%, 14.4%, 11.3%, and 16.5%, respectively. Approximately 70% of the patients in this study received a median of four courses of concurrent chemotherapy. However, the incidence of grade ≥ 3 HT was lower than that reported in most studies. This may be attributed to the pelvic functional bone marrow being protecting well from external irradiating, and patient’s tolerance to radiotherapy and chemotherapy was improved.

To the best of our knowledge, the time of HT occurrence during radiotherapy is rarely reported in the literature. We are the first to report that the median occurrence time of grade ≥ 2 leukopenia, neutropenia, thrombocytopenia and anemia was the 29th day, 42th day, 35th day, and 31th day, respectively. The median occurrence time of grade ≥ 3 leukopenia, neutropenia, thrombocytopenia and anemia was the 38th day, 40th day, 35th day, and 41.5th day, respectively. Preclinical studies have shown that when bone marrow is exposed to a large volume of 30–40 Gy radiation, neutropenia occurs after 1 week, thrombocytopenia occurs after 2–3 weeks and hemoglobin reduction occurs after 2–3 months, while irradiation with over 50 Gy may cause irreparable damage to the microcirculation of bone marrow [1]. Therefore, it is suggested that in the 4th week of radiotherapy, patients should be alert to the occurrence of grade ≥ 2 HT, in the 5th to the 6th weeks of radiotherapy, patients should be paid more attention to the occurrence of grade ≥ 3 HT, and supportive treatment should be taken when necessary.

Radiation can lead to HT [23, 24] by damaging the pelvic bone marrow hematopoietic stem cells and bone marrow mesenchymal stem cells and causing disorders of the vascular system and bone marrow microenvironment. Pelvic bone marrow-sparing intensity-modulated radiotherapy (PBMS-IMRT) can reduce the risk for HT during radiotherapy in patients with uterine cervical cancer [4, 7, 14]. However, the delineation methods for pelvic bone marrow and the recommended dose-volume parameters are still lacking. Therefore, it is particularly important to identify active bone marrow hematopoietic regions based on imaging. The INTERTECC study [14] confirmed that PET-CT-guided PBMS-IMRT can significantly reduce the risk of grade 3 neutropenia (8.6% vs. 27.1%, P = 0.035), but PET-CT has a high economic cost, so the clinical application of PET-CT has been limited.

The delineation methods for pelvic bone marrow vary greatly among different studies. Most studies adopted the definition of pelvic bone marrow recommended by Mell et al. [6], who delineated the outline of the pelvic bone within the radiotherapy target volume from the top of the lumbar vertebra to the bottom of the ischial tubercle (including the bilateral femoral heads and upper segments of the femur). Mell et al. [6] collected 37 cases of uterine cervical cancer patients receiving concurrent chemotherapy and radiotherapy, and the results showed that pelvic bone marrow V10 was significantly related to the occurrence of grade ≥ 2 leukopenia and neutropenia, and a pelvic bone marrow V10 < 90% was recommended. Albuquerque et al. [5] analyzed 40 uterine cervical cancer patients who received concurrent chemoradiotherapy, and the results showed that the total pelvic bone marrow V20 was significantly related to the occurrence of grade ≥ 2 HT; the total pelvic bone marrow V20 < 80% was recommended. Rose et al. [4] analyzed 44 cases of patients with uterine cervical cancer who received adjuvant IMRT and concurrent chemotherapy after surgery, and the results showed that V10 and V20 of the total pelvic bone marrow were significantly related to the occurrence of grade ≥ 3 HT, and V10 < 95% and V20 < 76% of the total pelvic bone marrow were recommended. Wang et al. [8] analyzed 232 cases of uterine cervical cancer patients who received radiotherapy (IMRT or TOMO) and concurrent chemotherapy, and univariate analysis showed that Dmax, Dmean, etc., of the total pelvic bone marrow were related to the occurrence of grade ≥ 2 HT, while multivariate analysis showed that V16 − 18, V35, V36, and V40 of the total pelvic bone marrow were significantly related to the occurrence of grade ≥ 2 HT. Additionally, the definition of pelvic bone marrow in the ROTG 0418 clinical trial [9] was similar to that of Mell et al. [6], who delineated all bones within the radiotherapy target volume, excluding the femoral neck and corpus femoris. This trial recruited 83 patients with gynecological tumors receiving adjuvant IMRT after surgery (40 uterine cervical cancer patients received radiotherapy and concurrent chemotherapy, and 43 endometrial cancer patients received radiotherapy alone). The results showed that the pelvic bone marrow V40 and Dmean were significantly related to the occurrence of grade ≥ 2 HT, and pelvic bone marrow V40 ≤ 37% and Dmean ≤ 34.2 Gy were recommended. Zhou et al. [10] collected 31 cases of uterine cervical cancer patients receiving concurrent chemoradiotherapy, and used FDG-PET/CT to identify active pelvic bone marrow (exceeding the systemic average SUV). The results showed that the active bone marrow V40 > 738 cc played a stronger role in predicting HT than values obtained through conventional pelvic bone marrow delineation.

In the multivariate regression analysis in this study, only the Dmax of FBM1, V10 of FBM1 and Dmean of FBM2 were respectively significantly associated with grade ≥ 2 leukopenia, grade ≥ 2 anemia, and grade ≥ 2 thrombocytopenia. We found that patients with FBM1 Dmax < 53 Gy, FBM2 Dmean < 33 Gy and FBM1 V10 < 95% had significantly lower the incidence of grade ≥ 2 HT. All these results were significantly different from those reported above, which can be mainly attributed to the outline of FBM1/2 and good dose constraints to the FBM1/2. Moreover, Dmax of FBM1 had an AUC value of 0.819 for predicting grade ≥ 2 leukopenia, with the sensitivity and specificity values of approximately 80%, when the cutoff value was 53 Gy, indicating that Dmax of FBM1 had good prediction efficiency. As most studies had shown, Dmax of pelvic bone marrow was not related with HT. Our results also showed that there was not a statistical relationship between the Dmax of FBM2 and grade ≥ 2 HT. However, there was a statistical relationship between the Dmax of FBM1 and grade ≥ 2 leukopenia, which may be mainly due to differences in delineation of pelvic bone marrow and high heterogeneity of bone marrow hematopoiesis. The high dose irradiation of FBM1, which had the strongest hematopoietic function, significantly limited the hematopoietic function.

However, according to the ROC curves of FBM2 Dmean for predicting the incidence of grade ≥ 2 thrombocytopenia and FBM1 V10 for predicting the incidence of grade ≥ 2 anemia, the AUC values, the sensitivities and specificities of cutoff values were unsatisfactory. The reason may be attributed to the facts that leukopenia is the first reaction of HT, followed by thrombocytopenia and anemia during chemoradiotherapy, and the FBM2 had inferior hematopoietic ability to FBM1, which may further lessen the predictive efficiency of the dose-volume parameters.

In addition, this study found that a boost dose to the lymph nodes/parametrial tissue was an independent risk factor for grade ≥ 2 leukopenia, neutropenia and thrombocytopenia. Compared with 45 Gy/25 F of external pelvic irradiation, the prescribed dose of 50.4 Gy/28 F led to a higher risk for grade ≥ 2 leukopenia and neutropenia, which was also consistent with the finding that the high-dose volume parameters were associated with HT. Therefore, radical radiotherapy, especially when in combination with a boost dose to the lymph nodes/parametrial tissue volume, not only leads to acute HT during radiotherapy but also leads to a certain risk of late bone injury such as incomplete pelvic fracture after radiotherapy [25]. It is also well known that concurrent chemotherapy is an important factor that causes HT. Keys et al. [26] reported that the incidence of grade 3 HT for uterine cervical cancer patients in the chemoradiotherapy group was significantly higher than that in the radiotherapy alone group (21% vs. 2%). This study likewise found that the courses of concurrent chemotherapy were significantly associated with grade ≥ 2 leukopenia, neutropenia and anemia. When patients receive chemotherapy before radiotherapy, this may theoretically accelerate HT during radiotherapy. However, multifactor analyses of our study did not find the relationship between grade ≥ 2 HT and the courses of induction chemotherapy before radiotherapy, this result may be due to the limited sample size and factors such as the dose-volume parameters of FBM1/2, concurrent chemotherapy, and boost doses to the lymph nodes/parametrial tissue. Moreover, 30-90% of cancer patients also had anemia, and its occurrence and severity is related to patient age and disease course [27]. This study found that old age and low body weight were high-risk factors for the occurrence of grade ≥ 2 anemia during radiotherapy for uterine cervical/endometrial cancer. Therefore, concurrent chemoradiotherapy for uterine cervical /endometrial cancer patients, especially those with boost doses to the lymph nodes/parametrial tissue, is associated with a high risk of HT, and patients should be closely monitored through blood counts during radiotherapy. Elderly patients with low body weight should be given more attention to remain highly vigilant of HT in clinical practice.

Functional bone marrow protection is valuable. Pelvic bone marrow is the main hematopoietic organ of adults, accounting for approximately 40% of the whole bone marrow [28]. Bone marrow mainly divides into red bone marrow with hematopoietic function and yellow bone marrow with a large amount of adipose tissue. At present, most studies have regarded the whole bone as a dose-limiting organ and discussed the relationship between bone marrow dose-volume and HT. However, an IMRT plan that strictly limits the dose volume of the whole pelvic bone marrow may affect the coverage of the target volume and the protection of other important OARs, such as the rectum and bladder. Therefore, dose constraints to bone marrow with active hematopoietic function can better reduce the risks for HT and minimize the bad impact on the target coverage and surrounding normal organs protection. Compared with delineating the whole pelvic bones, the delineation of functional bone marrow is more conducive to shortening doctors′ manual delineation time and improving clinical efficiency.

In addition, bone marrow is also the birthplace of all kinds of immune cells, the place where B lymphocytes mature and where the humoral immune response occurs. A meta-study suggested that grade 3 lymphocytopenia induced by radiotherapy was associated with poor prognosis [29]. Therefore, pelvic radiotherapy not only damages the hematopoietic system of the body, but also may have a certain inhibitory effect on the immune system. Dose constraints for the functional bone marrow may protect immune cells as much as possible and improve the tolerance of patients to radiotherapy or chemoradiotherapy. In the current era of immunotherapy, protecting immune cells may improve the efficacy of immunotherapy and may be the research direction of pelvic radiotherapy in the future. Although this study did not discuss the relationship between FBM and lymphocytopenia, it is worth further analysis.

There were some shortcomings in this study. First, this research method was based on the functional bone marrow defined by metabolic imaging, and there were no cellular biology experiments to verify the heterogeneity of pelvic bone marrow hematopoietic function. Second, this study was a single-center, small-sample, nonrandomized controlled study, and the conclusion of this study needs to be verified by more multicenter, prospective, randomized controlled studies. Third, we did not analyze the effect of brachytherapy on myelosuppression, though this effect may be very small.

In conclusion, for patients of uterine cervical/endometrial cancer receiving radiotherapy with or without concurrent chemotherapy, grade ≥ 2 HT usually occurs in the 4th week of radiotherapy and grade ≥ 3 HT usually occurs in the 5th to 6th week of radiotherapy. The Dmax, and V10 of FBM1 and the Dmean of FBM2 were significantly associated with the occurrence of grade ≥ 2 HT. More courses of concurrent chemotherapy, a higher prescribed dose of external pelvic irradiation and a boost dose to the lymph nodes/ parametrial tissue were independent factors for an increased risk of grade ≥ 2 HT. Older age and lower weight increased the risk of grade ≥ 2 anemia. The recommended optimal dose constraints were FBM1 Dmax < 53 Gy, V10 < 95%, and FBM2 Dmean < 33 Gy.