Evaluation of the rectal V30 parameter in patients diagnosed with postoperative endometrial cancer.

BACKGROUND: The present paper reports on analysis of 184 patients who were diagnosed with endometrial cancer. The main objective of this study was to address parameter Vrec(30Gy) which determines a volume of the rectum irradiated with a dose of 30 Gy during radiotherapy.
MATERIALS AND METHODS: All patients were irradiated with an IMRT technique on linear accelerators. The planning target volume (PTV) contour was determined by a radiation oncologist. The clinical target volume (CTV) was drawn on CT images obtained in a prone position. For statistical analysis, appropriate tests (e.g. the Shapiro-Wilk, Wilcoxon) were used. RESULTS AND DISCUSSION: The performed analysis showed that the recommended condition for Vrec(30Gy) is met only in 3% of patients and the observed median value exceeds 90%. The obtained results were compared with the studies in which the Vrec(30Gy) values were related to various radiotherapy techniques.
CONCLUSIONS: The analysis showed that the condition for Vrec(30Gy) is satisfied in the case of only 3% of patients. Due to the difficulty with meeting the condition, it should be reconsidered based on real results.

Introduction

Modern radiotherapy techniques allow a homogeneous dose distribution to be delivered to the tumor region while minimizing the dose to the surrounding healthy tissues. The approach to protect healthy tissues depends on the structure of the organs at risk (functional subunit organization) and on the ability to repair radiation damage and restore normal function after receiving a given dose of radiation [1]. Information about dose limits for a given organ at risk — dose (Gy) or dose-volume parameters and the associated risk (%) was provided by RTOG 0418 Quantec Reports. When preparing a treatment plan, this data help to define acceptable dose limits and estimate the risk of radiation-induced complications. Endometrial cancer is the sixth most common cancer in women and it constitutes 4.4% of all cases. In 2018, about 382,069 women were diagnosed with endometrial cancer [2]. The survival is relatively high — the 5-year relative survival is about 81% [3]. With greater numbers of long-term survivors, there is a growing need to reduce the toxicity of treatment to the organs at risk and develop better ways to measure and treat late-onset side effects [4]. In the case of postoperative endometrial cancer, one of the organs at risk is the rectum. The most commonly reported side effects occurring for the rectum after radiation therapy are defecation urgency, rectal bleeding and bloating [5, 6]. To avoid such adverse effects, the RTOG Reports recommend that the dose of 30 Gy should not exceed 60% of the rectum volume (Vrec(30Gy) < 60%). The RTOG restriction is difficult to be satisfied because of the concave-shaped target and the closeness of the rectum. The rectum is contained inside the concavity and very often has a common part with the irradiated area. During the evaluation of a treatment plan, usually two solutions are considered: acceptance of the dose distribution with an increased probability of rectal complications or sparing of the rectum with RTOG guidelines with a decreased dose in the tumor area.

Materials and methods

The retrospective study involved all 184 patients (2013–2018) treated for postoperative endometrial cancer in the West Pomeranian Oncology Center in Szczecin, Poland. The patients underwent intensity-modulated radiation therapy (IMRT) that included a therapeutic dose of 45 Gy in 25 fractions. Each planning target volume (PTV) contour was determined and constructed by a radiation oncologist following the institutional protocol. The clinical target volume (CTV) was drawn on CT images obtained in a prone position with a full bladder (at least 0.5 hour after drinking 0.7 dm3 of water) and the empty rectum. Those CT images were fused with images performed after emptying the bladder. The CTV contour consisted of the ITV covering the parametrial and vaginal cuff (drawn on CT with a full and empty bladder) and the nodal CTV (drawn on CT with a full bladder) covering the internal, external, common iliac, obturator and presacral lymph nodes on both sides. Than the margin to generate PTV was added: 5 mm for the ITV and 8 mm for the nodal CTV. Treatment was planned in the Prowess Panther (Radiology Oncology Systems, Inc., San Diego, CA, USA), version 5.10, using 9–10 6 MeV photon beams and it was assumed that over 98% of the PTV should receive at least 95% of assigned dose, and no more than 2% of the PTV should exceed 107% of the assigned dose. The acceptable rectal dose was as that recommended by the Radiation Therapy Oncology Group (RTOG) 0418 — rectum volume < 60% to receive a dose 30 Gy. All patients were irradiated with a step-and-shoot IMRT technique on Siemens linear accelerators (Siemens Healthcare, Erlangen, Germany).

Figure 1 shows the clinical volumes used in the statistical analysis. For the numerical purposes, the Python language together with SciPy, Statistics and Matplotlib modules were used [7, 8]. The dimensionless parameter describing the ratio of the rectum volume in a target and the total rectum volume Vrec(tar)/Vrec(tot) was introduced and used in the analysis. The normality of the data distribution and correlations between particular parameters were examined using the Shapiro-Wilk test and the Spearman correlation analysis, respectively. The analyzed data were compared with the results obtained in several papers [9-18] in which different types of radiation therapy used in the patient treatment were described. The statistical significance of differences between calculated values was verified by the one-sample Wilcoxon non-parametric hypothesis test. The examined Vrec(30Gy) data were compared with respective Vrec(30Gy) values from the mentioned papers. The objective of this prospective study was to evaluate the parameter Vrec(30Gy) and its correlation with the naturally fixed parameters such as: rectum volume in a target (Vrec(tar)), total rectum volume (Vrec(tot)), planned target volume (VPTV) and ratio of the rectum volume in target and total rectum volume (Vrec(tar)/Vrec(tot)). An important part of the proposed study was also the comparison of the obtained data with those from other clinical trials. Finally, the last purpose of the presented research was to create an additional statistical tool to evaluate and predict the dose distribution in the rectum by relating the obtained results to the population of patients irradiated due to postoperative endometrial cancer.

Figure 1

Clinical volumes: rectum volume in a target (Vrec(tar)) outlined in blue hatched area; total rectum volume (Vrec(tot)) outlined in green; planned target volume (VPTV) outlined in red

Results and Discussion

The Shapiro-Wilk test showed that most of the data considered are not normally distributed (Tab. 1). For this reason the correlation analysis was performed using the Spearman method.

Table 1

The Shapiro-Wilk test results for the data considered

Parameter	p-value
Vrec(30Gy) (%)	< 0.01*
Vrec(tar) [cc]	< 0.01*
Vrec(tot) [cc]	< 0.01*
VPTV [cc]	0.52
Vrec(tar)/Vrec(tot)	< 0.01*

Significance level = 0.05. Particular symbols denote: Vrec(30Gy) — rectum volume (in %) with 30 Gy dose, Vrec(tar) — rectum volume in a target, Vrec(tot) — total rectum volume, VPTV — planned target volume, Vrec(tar)/Vrec(tot) — ratio of the rectum volume in a target and total rectum volume.

significant results

Table 2 gives the exact values of the Spearman coefficient and the p-value for particular data. It is noticeable that the strongest linear correlation occurs between Vrec(30Gy) and Vrec(tar)/Vrec(tot), VPTV and Vrec(tar)/Vrec(tot) as well as in case Vrec(tar) and Vrec(tot) (however, a relatively strong correlation between Vrec(tot) and Vrec(tar) seems to be obvious). The correlation analysis shows that there is no linear relationship between Vrec(tot) and other parameters, or it is very weak. It is worth noting that the introduced parameter Vrec(tar)/Vrec(tot) has a linear relationship mostly with Vrec(tar) (a linear correlation of Vrec(tot) and Vrec(tar)/Vrec(tot) should be considered as irrelevant).

Table 2

Results of the Spearman correlation analysis

		Vrec(30Gy)	Vrec(tar)	Vrec(tot)	VPTV	Vrec(tar)/Vrec(tot)
Vrec(30Gy)	Spearman coeff.	–	−0.21	−0.19	0.21	0.48
	p-value	–	< 0.01*	0.01*	< 0.01*	< 0.01*
Vrec(tar)	Spearman coeff.	0.21	–	0.65	0.37	0.71
	p-value	< 0.01*	–	< 0.01*	< 0.01*	< 0.01*
Vrec(tot)	Spearman coeff.	−0.19	0.65	–	0.06	−0.01
	p-value	0.01*	< 0.01*	–
VPTV	Spearman coeff.	0.21	0.37	0.06	–	0.48
	p-value				–
Vrec(tar)/Vrec(tot)	Spearman coeff.	0.48	0.71	−0.01	0.48	–
	p-value	< 0.01*	< 0.01*	0.91	< 0.01*	–

significant results (significance level = 0.05)

Parameter of interest — Vrec(30Gy)

One of the objectives of this study was to examine how Vrec(30Gy) changes due to the naturally fixed parameters. Figures 2–5 show relations between the parameter of interest and other parameters. It is noticeable that most of the patients are generally distributed in the closeness of the highest Vrec(30Gy) value. From a medical point of view, this is not a desirable result - in paper [9] it is strongly recommended that the limit: „Rectum < 60% to receive ≥ 30 Gy” be kept — nevertheless, this is the case.

Figure 2

Patient distribution: Vrec(30Gy) – Vrec(tar)

Figure 3

Patient distribution: Vrec(30Gy) – Vrec(tot)

Figure 4

Patient distribution: Vrec(30Gy) – Vrec(tar)/Vrec(tot)

Figure 5

Patient distribution: Vrec(30Gy) – VPTV

Vrec(30Gy) comparison

Because of the non-normal distribution of data the median was used as a measure of the central tendency. The obtained median Vrec(30Gy) value is 93.6%. It means that the previously mentioned condition „Rectum < 60% to receive ≥ 30 Gy” is definitely not satisfied.

In this subsection, we tried to show how difficult it is to satisfy the condition presented in paper [9], in spite of the fact that in the RTOG 0418 study patients underwent postoperative IMRT with prescribed dose of 50.4 Gy in 28 fractions. The analyzed data were compared with the results obtained in several papers [9-18] in which different types of radiation therapy used in the patient treatment were described.

The examined data for Vrec(30Gy) were compared with respective Vrec(30Gy) values from the mentioned papers. In Table 3, the results of the Wilcoxon test are presented.

Table 3

One-sample Wilcoxon test results

Vrec(30Gy) [%]	p-value	Source	Comments	Dose [Gy]/fractions
60.0	< 0.01*	[9]	IMRT	50.4/28
65.1	< 0.01*	[10]	IMRT	–
82.2	< 0.01*	[11]	HT	50.4/28
98.4	< 0.01*	[11]	CT
97.9	< 0.01*	[15]	CT	47.8/–
94.5	< 0.05*	[15]	IMRT	47.4/–
97.7	< 0.01*	[16]	CRT	45.0/25
97.4	< 0.01*	[16]	FIF
81.9	< 0.01*	[17]	Halcyon (node positive patients)	45.0–55.0/25
89.7	0.5	[17]	HT (node positive patients)
92.4	0.07	[17]	Halcyon (node negative patients)
97.4	< 0.01*	[17]	HT (node negative patients)

IMRT — intensity-modulated radiation therapy; HT — helical tomotherapy; CT — conventional therapy; CRT — conformal radiotherapy; FIF — field-in-field technique;

significant results (significance level = 0.05)

First, the significance of the difference between the condition from paper [9] and the analyzed Vrec(30Gy) data was examined. It is not surprising that the mathematical difference between the condition Vrec(30Gy) < 60% and the obtained median Vrec(30Gy) = 93.6% value is also statistically significant (Tab. 3). A comparable value, Vrec(30Gy) = 65.1%, is presented in paper [10], in which twenty-eight patients receiving postoperative IMRT for gynecological malignancies were examined. From the statistical point of view, a difference between the Vrec(30Gy) value presented in [10] and the obtained median is indisputably significant.

Paper [11], in which 10 histologically confirmed cervical cancer patients were enrolled, shows that the use of helical tomotherapy (HT) significantly decreases the average percentage of the rectum dose-volume (compared to the conventional radiotherapy) (see Tab. 3). A very similar conclusion was made in papers [12, 14] — Vrec(30Gy) ~ 87–90% in the case of using IMRT techniques for gynecologic treatment. Nevertheless, the presented values definitely differ from the condition Vrec(30Gy) < 60% from paper [9]. Although statistical comparison between Vrec(30Gy) presented in [11] (in the case of the conventional and IMRT techniques) and the analyzed data showed significant differences (Tab. 3), both can be considered to be at similar level. To conclude, in paper [11] (as well in [12-14]) the same „problem” of relatively large values of the parameter Vrec(30Gy) also occurred.

The ability of IMRT to reduce the volume of small bowel irradiated in women with gynecologic malignancies receiving whole pelvic radiotherapy has been investigated by in a paper by Roeske et al. [15]. The analysis was carried out on a group of ten women with cervical or endometrial cancer. The effect of radiotherapy on the bladder and rectum was taken into account. As a result, the average rectum dose-volume for 30 Gy decreased by 3.5%. Even greater volume reductions were observed for 40 and 45 Gy, 19.6% and 28%, respectively. The parameter Vrec(30Gy) for the conventional and IMRT techniques is presented in Tab. 3. The statistical analysis showed significant differences between these values and the considered data. Nevertheless, the reported values are greater than the criterion Vrec(30Gy) < 60%.

In paper [16] an attempt was made to compare two techniques of radiotherapy in terms of dosimetric benefits for early stage endometrial cancer patients. The study concerned the conformal radiotherapy (CRT) and the field-in-field technique (FIF). Ten patients were included in the study for whom two treatment plans were created. In the case of the FIF technique, a reduction in radiation doses was observed for most parameters, e.g. PTV. The biggest difference was observed especially for doses > 45 Gy — the irradiated volumes of OAR (organs at risk) were significantly reduced with the FIF technique. For Vrec(30Gy) however, only the bowel irradiated volume was significantly reduced. The rectum volume irradiated by 30 Gy changed from 97.4 (FIF) to 97.7 (CRT). When compared to the analyzed data, a significant difference for the rectum volume was observed for both FIF and CRT results (see Tab. 3).

A comparison of volumetric modulated arc therapy employing a dual-layer stacked multi-leaf collimator and helical tomotherapy for cervix uteri cancer has been presented in paper [17]. A retrospective treatment planning study was performed on a cohort of 20 patients. Ten cases were selected with positive lymph nodes and 10 without positive nodes. The retrospective treatment planning study suggested the essential equivalence between Halcyon based [18] and helical tomotherapy based plans. In fact, even with the new Halcyon radiotherapy platform, the achieved Vrec(30Gy) value exceeds 80% and 90% for node positive and node negative patients, respectively. The statistical comparison is presented in Table 3.

Figures 6 and 7 show histograms related to Vrec(tar)/Vrec(tot) and Vrec(30Gy), respectively. Additional (red and blue) lines (Fig. 7) indicate the first decile and median values for the considered parameter. The quantile analysis showed exactly the small number of patients for whom the Vrec(30Gy) value satisfied the condition presented in [9], i.e. Vrec(30Gy) < 60% (see Tab. 4). In fact, the condition Vrec(30Gy) = 60% is the third percentile of the analyzed data.

Figure 6

Histogram and patient density distribution function: Vrec(tar)/Vrec(tot)

Figure 7

Histogram and patient density distribution function: Vrec(30Gy)

Table 4

Results of the quantile analysis — decile values

Deciles	Vrec(30Gy)
1	70.2
2	79.0
3	84.1
4	89.8
5	93.6
6	96.4
7	99.0
8	100.0
9	100.0

Conclusions

The main objective of this study was to address the Vrec(30Gy) which determines a volume of the particular organ at risk (i.e. the rectum) irradiated with a dose of 30 Gy during radiotherapy. The statistical analysis mainly included the correlation analysis between Vrec(30Gy) and other collected parameters. The strongest correlation was observed in the case of the introduced parameter Vrec(tar)/Vrec(tot).

The second part of the study concerned the problem of compliance with the condition from paper [9] - Vrec(30Gy) < 60%. The presented analysis showed that the condition was satisfied in the case of only 3% of patients. The observed median Vrec(30Gy) value was 93.6% and it is comparable, in mathematical sense, with the values achieved in many papers, e.g. [11-17]. Definitely, the problem with meeting the condition Vrec(30Gy) < 60% is not related to the radiotherapy technique. The use of relatively new techniques of radiotherapy did not solve this „problem” either. In our opinion, the condition Vrec(30Gy) < 60% from paper [9] should be reconsidered based on real results obtained in different radiotherapy centers.

Because of the issues mentioned above, it is difficult to propose a universal statistical tool to evaluate and predict a dose in the rectum area. Despite the significant linear correlation between Vrec(30Gy) and Vrec(tar)/Vrec(tot), it is not possible to predict the rectal dose distribution based on the introduced ratio parameter Vrec(tar)/Vrec(tot).To develop a statistical tool for predicting the dose distribution in the rectum, more information should be taken into account (e.g. patient anatomy, age, etc.) — this will be the objective of our further research.