Predicting tumor recurrence in patients with cervical carcinoma treated with definitive chemoradiotherapy: value of quantitative histogram analysis on diffusion-weighted MR images

Abstract

Background

Further research is required for evaluating the use of ADC histogram analysis in more advanced stages of cervical cancer treated with definitive chemoradiotherapy (CRT).

Purpose

To investigate the utility of apparent diffusion coefficient (ADC) histogram derived from diffusion-weighted magnetic resonance images in cervical cancer patients treated with definitive CRT.

Material and Methods

The clinical and radiological data of 50 patients with histologically proven cervical squamous cell carcinoma treated with definitive CRT were retrospectively analyzed. The impact of clinicopathological factors and ADC histogram parameters on prognostic factors and treatment outcomes was assessed.

Results

The mean and median ADC values for the cohort were 1.043 ± 0.135 × 10⁻³ mm²/s and 1.018 × 10⁻³ mm²/s (range, 0.787–1.443 × 10⁻³ mm²/s). The mean ADC was significantly lower for patients with advanced stage (≥IIB) or lymph node metastasis compared with patients with stage <IIB or no lymph node metastasis. The mean ADC, 75th percentile ADC (ADC75), 90th percentile ADC (ADC90), and 95th percentile ADC (ADC95) were significantly lower in patients with tumor recurrence compared with patients without recurrence. In multivariate analysis, tumor size, ADC75 and ADC95 were independent prognostic factors for both overall survival and disease-free survival.

Conclusion

ADC histogram parameters could be markers for disease recurrence and for predicting survival outcomes. ADC75, ADC90, and ADC95 of the primary tumor were significant predictors of disease recurrence in cervical cancer patients treated with definitive CRT.

Keywords

Cervical cancer magnetic resonance imaging (MRI)apparent diffusion coefficient chemoradiotherapy prognostic factor

Introduction

Cervical cancer is the second most common cancer in women worldwide, with almost 80% of cases arising in low-income countries according to the World Health Organization (WHO) (1,2). The most important prognostic factors in cervical cancer are stage at diagnosis, tumor diameter, histological subtype, lymph node metastases, and lymphovascular space invasion (3). For locally advanced cervical cancer, radiotherapy (RT) with concurrent cisplatin-based chemotherapy has been widely accepted (4,5). However, most patients with high-risk features recur locally with distant metastasis. For this reason, if patients at high risk for disease recurrence after definitive chemoradiotherapy (CRT) could be reliably identified, better treatment could be accomplished by conducting more intensive follow-ups (6).

Diffusion-weighted magnetic resonance imaging (DWI-MRI) can non-invasively characterize alteration of tissue cellularity and integrity of cellular membranes by evaluating diffusion properties of water molecules. This measurement allows for a comparison of the diffusion properties of cervical cancer tissue and healthy cervical tissue (7,8). DWI has been applied to detect and characterize tumors using an apparent diffusion coefficient (ADC) value. Several studies have confirmed that the mean ADC values for cervical cancer are significantly lower than those for normal cervix; however, the reported ADC values for cervical cancer and normal cervix vary considerably (9,10). DWI has a potential value in the assessment of pathologic features of cervical cancer. Previous studies suggested that ADC values may provide useful information regarding tumor cellularity, tumor aggressiveness, and subtype characterization (10 –13). It has been previously documented that ADCs of well or moderately differentiated cervical cancer are significantly higher than those of poorly differentiated cervical cancer (10,13). Also, significant correlations between the ADC of the primary cervical tumor and clinicopathologic characteristics and treatment outcomes in patients treated with surgery and definitive CRT was demonstrated (14,15). However, the distribution of ADC values varies within any tumor; therefore, the mean ADC may not represent the full spectrum of histology within a tumor (16). Methods of ADC histogram analysis different from conventional region of interest (ROI)-based measurements and able to represent the whole solid viable component of the neoplasm might overcome the above mismatching results. An ADC histogram can display ADC values and distribution within a whole tumor and analyze ADC voxel by voxel, thereby providing more information than the mean ADC. It has been previously reported that pretreatment ADC histogram analysis may serve as a biomarker for predicting tumor recurrence in patients with cervical cancer treated with either surgery or definitive CRT (11,17 –19). Recently, two studies showed that the quantitative histogram analysis of an ADC map helped to characterize the histologic feature of early stage cervical cancer that are indicative of poor prognosis (18,19). However, most of the studies had a limited number of patients and these were mostly with early stage disease. Further research is required for evaluating the use of ADC histogram analysis in more advanced stages of cervical cancer treated with definitive CRT. The purpose of the current study is to investigate the utility of the ADC histogram derived from DWI in cervical cancer patients treated with definitive CRT.

Material and Methods

Patients

From February 2011 to June 2014, 225 consecutive women with uterine cervical cancer diagnosed by biopsy underwent pre-treatment baseline pelvic MRI or positron emission tomography (PET-CT). Of these, 150 were excluded from this study, because only pretreatment PET-CT was performed in these patients. Four patients whose tumors were not visible on MRI, four patients with adenocarcinoma histology, and 17 patients who received prior chemotherapy were also excluded from the patient cohort. Thus, the clinical and radiological data of 50 patients with histologically proven cervical squamous cell carcinoma treated with definitive CRT were retrospectively analyzed. This study was approved by Baskent University Institutional Review Board (Project no: KA 15/147) and supported by Baskent University Research Fund.

Treatment

Patients without distant metastases were treated with a combination of 3D conformal external beam RT (3DCRT) with concurrent weekly 40 mg/m² cisplatin and high-dose-rate brachytherapy (HDR BRT) as described previously (20). In total, 50.4 Gy external RT (1.8 Gy per fraction, daily, Monday through Friday) was delivered using 18-MV photons. Three-dimensional brachytherapy planning was performed using 7 Gy per fraction prescribed to the target minimum, given in four fractions.

ADC histogram analysis

The ADC map was constructed automatically on a pixel-by-pixel basis using the standard software on the console (Syngo, Siemens, Erlangen, Germany). ADC values were measured by one radiologist with more than 10 years of experience in pelvic MRI (EK), and checked by other experienced radiologists (GE and ZK) who were blinded to clinical and pathological data. The observer evaluated both axial DW image and ADC map side-by-side and used a spatial cursor key to match the section level between axial DW image and ADC map. The tumor border was identified by visual evaluation on axial DW image. The radiologist manually outlined a ROI on the pre-contrast image and on one image from post-contrast phase. The ROI was extended as far as possible into the solid components of the tumor while avoiding the areas of reduced signal intensity corresponding to necrosis and ischemia within the tumor (Fig. 1). After ROI definition, an ADC histogram of each section was generated by the software. After the summation of pixel ADCs from all the slices that showed the tumor, several parameters were calculated through histogram analysis. The mean ROI of the tumor on ADC maps was 812 mm² (range, 256–2352 mm²).

Fig. 1.

A 61-year-old FIGO stage IIB cervical cancer MRI (a) demonstrating hyper-intense lesion with diffusion-weighted at b-800 images. (b) In the axial plane, tumor borders are drawn manually for ADC histogram analysis.

Clinical follow-up

Clinical follow-up of patients was performed every 3 months for 2 years, then every 6 months up to 5 years and annually thereafter. At least 3 months after the completion of treatment, all patients received MRI or PET-CT were taken in suspected cases. Biopsy was performed in suspected cases. Biopsy was not performed before 6 months after completion of CRT. Failure was defined as biopsy-proven recurrence or documented progression of disease in serial-imaging studies (21).

Statistical analysis

Statistical analyses were performed using SPSS software version 20.0 (SPSS Inc., Chicago, IL, USA). The time to event was calculated as the time interval from the date of diagnosis to the date of first finding on clinical or imaging examination that suggested disease recurrence. The ADC histogram of every sagittal section was converted into a corresponding frequency table using commercially available software (Matlab 2011). ADC_min is the lowest ADC of one voxel within the ROI, ADCn% is the threshold value at which n% of the voxel values that form the histogram have a lower value (22). ADC_mean is the average of all the ADC values within the ROI. The parameters derived from the ADC histogram are as follows: mean; 2.5th, 5th, 10th, 25th, 50th, 75th, 90th, 95^th, and 97.5th percentiles. The histogram parameters were compared between patients with or without tumor recurrence after definitive CRT using Student's t-test or the Chi-square test.

Univariate and multivariate analyses of potential prognostic factors were performed using the Cox proportional hazards regression model. A receiver operating characteristic (ROC) curve was used to determine the most clinically useful cutoff value of variables in predicting tumor recurrence. The differences in overall survival (OS) and disease-free survival (DFS) rates according to the cutoff ADC value were evaluated using the Kaplan–Meier method with a log-rank test. A level of P < 0.05 was considered statistically significant.

Results

Patient characteristics and outcomes

Patient characteristics are presented in Table 1. The median age was 56 years (age range, 21–85 years). Mean tumor size was 5.3 ± 1.7 cm. Most of the patients (76%) had locally advanced stage (FIGO stage ≥ IIB) disease and all patients had squamous cell carcinoma. All patients received at least four cycles of weekly cisplatin-based concurrent chemotherapy.

Table 1.

Patient and tumor characteristics.

Characteristics	n (%)
Age (years) (median, range)	56 (21–85)
Tumor size (cm) (mean ± SD)	5.3 ± 1.7
FIGO Stage
IB	12 (24)
IIA	0 (0)
IIB	19 (38)
IIIA	6 (12)
IIIB	6 (12)
IVA	7 (14)
Pelvic lymph node metastasis
No	26 (52)
Yes	24 (48)

The median follow-up for all patients and surviving patients was 44 months (range, 3–86 months) and 52 months (range, 6–84 months), respectively. Of the 50 patients in the study cohort, 22 (50%) developed local, loco-regional, or distant failure. Of these, nine (18%) developed distant metastases, eight (16%) had pelvic recurrence, and five (10%) developed both pelvic and distant failure. At the time of the last follow-up, 29 patients (58%) were alive, two of the living patients are living with disease (4%), and 21 patients (42%) had died. Of these latter patients, 20 (40%) died due to disease and one (2%) died from other causes.

ADC analysis

The mean and median ADC values for the cohort were 1.043 ± 0.135 × 10⁻³ mm²/s and 1.018 × 10⁻³ mm²/s (range, 0.787–1.443 × 10⁻³ mm²/s). Mean ADC was significantly lower for patients with advanced stage (≥IIB) disease as compared with patients with early stage (<IIB) disease (1.019 ± 0.109 × 10⁻³ mm²/s versus 1.118 ± 0.182 × 10⁻³ mm²/s; P = 0.03) (Table 2). Mean ADC was significantly lower in patients with lymph node metastasis compared with patients without lymph node metastasis (0.928 ± 0.84 × 10⁻³ mm²/s versus 1.157 ± 0.170 × 10⁻³ mm²/s; P = 0.004).

Table 2.

Mean ± SD ADC values according to patients with early stage (<IIB) or advanced stage (≥IIB) disease.

ADC parameters	Stage < IIB (×10⁻³ mm²/s)	Stage ≥ IIB (×10⁻³ mm²/s)	P
ADCmean	1.118 ± 0.182	1.020 ± 0.109	0.03
ADCmedian	1.138 ± 0.184	1.011 ± 0.185	0.04
ADC2.5	0.451 ± 0.152	0.463 ± 0.170	0.82
ADC5	0.516 ± 0.145	0.535 ± 0.159	0.73
ADC10	0.603 ± 0.143	0.621 ± 0.151	0.71
ADC25	0.804 ± 0.155	0.784 ± 0.148	0.69
ADC50	1.118 ± 0.182	1.020 ± 0.109	0.03
ADC75	1.535 ± 0.215	1.291 ± 0.122	<0.001
ADC90	1.695 ± 0.213	1.491 ± 0.129	<0.001
ADC95	1.822 ± 0.212	1.611 ± 0.156	0.01
ADC97.5	1.875 ± 0.260	1.712 ± 0.224	0.07

In histogram analysis, ADC75, ADC90, and ADC95 were also significantly higher in patients with early stage disease compared with patients with advanced stage disease. As demonstrated in Table 3, the mean ADC, ADC75, ADC90, and ADC95 were significantly lower in patients with tumor recurrence compared with patients without recurrence. In addition, the difference between recurrence and no recurrence in ADC97.5 had borderline significance.

Table 3.

Mean ± SD ADC values of the primary tumor for recurrence and no recurrence groups.

ADC parameters	Recurrence	No recurrence	P
ADCmean	0.998 ± 0.073	1.078 ± 0.016	0.02
ADCmedian	1.015 ± 0.072	1.062 ± 0.247	0.39
ADC2.5	0.476 ± 0.148	0.447 ± 0.177	0.54
ADC5	0.539 ± 0.252	0.524 ± 0.170	0.73
ADC10	0.621 ± 0.129	0.612 ± 0.164	0.83
ADC25	0.794 ± 0.195	0.784 ± 0.164	0.82
ADC50	1.038 ± 0.096	1.071 ± 0.162	0.41
ADC75	1.269 ± 0.100	1.412 ± 0.205	0.004
ADC90	1.475 ± 0.110	1.590 ± 0.200	0.02
ADC95	1.527 ± 0.334	1.713 ± 0.213	0.02
ADC97.5	1.677 ± 0.213	1.808 ± 0.249	0.06

Prognostic factors for OS and DFS

Univariate analysis evaluating the clinical and ADC parameters demonstrated that tumor size, lymph node metastasis, and ADC75, ADC90, and ADC95 of the tumor were significantly associated for OS (Table 4). Tumor size, lymph node metastasis, ADC50, ADC75, and ADC95 were significant prognostic factors for DFS in univariate analysis. Using multivariate analysis, tumor size, ADC75, and ADC95 were independent prognostic factors for both OS and DFS.

Table 4.

Univariate and multivariate analysis for predicting overall survival and disease-free survival.

	Overall survival				Disease-free survival
Variables	Univariate analysis		Multivariate analysis		Univariate analysis		Multivariate analysis
Variables	HR (95% CI)	P	HR (95% CI)	P	HR (95% CI)	P	HR (95% CI)	P
Age	1.009 (0.973–1.046)*	0.64			1.000 (0.968–1.034)*	0.98
Tumor size	1.527 (1.171–1.990)*	0.002	1.577 (1.137–2.189)*	0.006	1.492 (1.152–1.932)*	0.002	1.632 (1.197–2.225)*	0.002
Stage
<IIB	1	0.14			1	0.21
≥IIB	2.557 (0.746–8.761)				2.568 (0.778–7.876)
Pelvic LN metastasis
Absent	1	0.007	1	0.27	1	0.002	1	0.12
Present	3.738 (1.428–9.782)		1.808 (0.627 – 5.211)		4.412 (1.706–11.410)		2.194 (0.795–6.055)
ADC values
ADCmean	1.005 (0.982–1.008)†	0.43			0.996 (0.985–1.005)†	0.53
ADCmedian	1.002 (0.991–1.013)†	0.76			1.001 (0.992–1.011)†	0.76
ADC2.5	1.030 (0.991–1.071)†	0.14			1.031 (0.996–1.067)†	0.09	0.999 (0.995–1.002)†	0.49
ADC5	1.029 (0.915–1.031)†	0.33			1.025 (0.924–1.030)†	0.37
ADC10	1.026 (0.917–1.034)†	0.39			1.030 (0.968–1.089)†	0.34
ADC25	1.029 (0.996–1.064)†	0.09	1.004 (1.000 – 1.009)†	0.06	1.025 (0.991–1.060)†	0.15
ADC50	1.008 (0.995–1.020)†	0.24			1.014 (1.000–1.028)†	0.04	0.999 (0.997–1.000)†	0.14
ADC75	1.024 (1.009–1.039)†	0.002	1.017 (1.001 – 1.032)†	0.004	1.019 (1.005 – 1.033)†	0.006	1.012 (1.003–1.021)†	0.01
ADC90	1.017 (1.004–1.020)†	0.02	1.018 (0.998 – 1.039)†	0.07	1.009 (0.995 – 1.024)†	0.21
ADC95	1.003 (1.001–1.005)†	0.01	1.002 (1.000 – 1.003)†	0.03	1.002 (1.000 – 1.004)†	0.02	1.011 (1.000–1.022)†	0.04
ADC97.5	1.004 (0.999–1.010)†	0.1	1.006 (1.000 – 1.012)†	0.10	0.998 (0.993 – 1.003)†	0.48

for increase of 0.01 standard deviation.

†

For increase of 1 standard deviation.

The relationship between ADC75 and ADC95 of the primary tumor and recurrence and survival was evaluated based on the cutoff value determined using ROC curve analysis. Fig. 2a shows the ROC curve analysis of the ADC75 and ADC95 of the primary tumor with respect to OS. The AUCs for ADC75 and ADC95 were 0.734 (95% CI, 0.595–0.873; P = 0.005) and 0.722 (95% CI, 0.581–0.864; P = 0.008), respectively. The cutoff values for ADC75 and ADC95 for OS were 1.289 × 10⁻³ mm²/s (sensitivity 69.0%, specificity 72.4%) and 1.626 × 10⁻³ mm²/s (sensitivity 69.0%, specificity 71.9%). As demonstrated in Fig. 2b, the AUCs for ADC75 and ADC95 were 0.729 (95% CI, 0.586–0.872; P = 0.006) and 0.692 (95% CI, 0.542–0.841; P = 0.02), respectively, for disease recurrence. The cutoff values for ADC75 and ADC95 for recurrence were 1.289 × 10⁻³ mm²/s (sensitivity 71.4%, specificity 72.4%) and 1.628 × 10⁻³ mm²/s (sensitivity 67.9%, specificity 68.2%).

Fig. 2.

The ROC analysis of ADC75 and ADC95 of the primary tumor for predicting (a) overall survival and (b) disease-free survival.

The 2-year OS and DFS rates for ADC75 > 1.289 × 10⁻³ mm²/s were significantly higher than those of patients with ADC75 < 1.289 × 10⁻³ mm²/s (82% versus 65%; P = 0.002 and 81% versus 44%; P = 0.006) (Fig. 3a and b). Similarly, the 2-year OS and DFS rates for ADC95 > 1.626 × 10⁻³ mm²/s were significantly higher than those of patients with ADC95 < 1.626 × 10⁻³ mm²/s (82% versus 62%; P = 0.006 and 79% versus 43%; P = 0.02) (Fig. 3c and d).

Fig. 3.

Overall survival curves of patients with ADC75 < 1.289 × 10⁻³ mm²/s and ≥1.289 × 10⁻³ mm²/s (a), and ADC95 < 1.626 × 10⁻³ mm²/s and ≥1.626 × 10⁻³ mm²/s (c). Disease-free survival curves of patients with pre-treatment ADC < 1.289 × 10⁻³ mm²/s and ≥1.289 × 10⁻³ mm²/s (b), post-treatment ADC < 1.626 × 10⁻³ mm²/s and ≥1.626 × 10⁻³ mm²/s (d), and ADC change <32.8% and ≥32.8% (d).

Discussion

In this study, we demonstrated that ADC75, ADC90, and ADC95 of the primary tumor were significant predictors of disease recurrence in cervical cancer patients treated with definitive CRT. Additionally, besides larger primary tumors, lower ADC75 and ADC95 were independent significant prognostic factors for OS and DFS. These findings support the evidence that, rather than mean and minimum ADC values, ADC histogram parameters could be used as markers for disease recurrence and for predicting survival outcomes.

Studies evaluating the ADC values in cervical cancer have been widely published (9,13,23). However, those studies were generally based on measurements based on local ROI, so the reported ADC values for cervical cancer and normal cervix varied considerably, which may in part be due to bias of ROI placement (9,16,24). Calculation of ADC values by defining ROIs on those voxels could result in false-negative results. Therefore, ADC evaluation based on the entire tumor is highly recommended. It was previously demonstrated that the ADC values in patients with squamous cell carcinoma were significantly lower than those in patients with adenocarcinoma (11,18). In a prospective study of 35 patients with early-stage cervical cancer and 38 controls on a 3 T, Lin et al. (18) concluded that the distribution of ADC assessed by histogram analysis helps in differentiating early-stage cervical cancer from normal cervix or benign lesions and evaluating the histologic features of neoplasms. Xue et al. (11) retrospectively analyzed the value of preoperative ADC histogram in assessing local aggressiveness of cervical cancer in 53 patients. The authors reported that ADC_mean, and ADC_min were significantly higher for adenocarcinoma of the cervix than for squamous cell carcinoma. Additionally, ADC_mean and ADC5–85 were significantly higher for well to moderately differentiated tumors than for poorly differentiated neoplasms. In another study of on a 1.5 T system, Downey et al. (19) found that the ADC_median was lower in poorly differentiated tumors compared to those of well or moderate differentiation in 60 stage I cervical cancer patients. It is difficult to make a detailed histopathological evaluation in patients treated with definitive CRT. An auxiliary non-invasive marker to predict treatment outcomes is essentially required for cervical cancer patients treated with definitive CRT. To homogenize our cohort in this study, we only evaluated patients with squamous cell carcinoma. The optimal parameter of the histogram for clinical assessment remains to be determined. Thus, the ADC histogram may be a promising tool for early assessment of treatment response for cervical cancer.

There are some studies evaluating the feasibility of ADC histogram in cervical cancer patients (11,17,18). In a study on a 3T-MRI, Heo et al. (17) reported that ADC histogram analysis might serve as a biomarker for predicting tumor recurrence in 42 women with histologically proven squamous cell carcinoma of the uterine cervix (>70% had advanced-stage disease) treated with definitive CRT. The authors found that the ADC_mean and ADC75 were higher in the recurrence group. Additionally, ADC75 was found to be a significant predictor of tumor recurrence in the multivariate analysis. In our study, rather than analyzing the role of ADC histogram parameters on recurrence, we also assessed the impact of ADC histogram on survival. We found that tumor size, ADC75 and ADC95 were significant prognostic factors for OS and DFS. We also analyzed the significant importance of ADC histogram parameters for defining high-risk and low-risk patients. We found that ADC75, ADC90, and ADC95 were significantly higher in patients with early stage disease compared with those with advanced stage disease.

Our study has limitations. First, we did not use multiple b values to calculate ADCs, using dual b values of 0 and 800 s/mm² instead. In female pelvic DWI, a higher b value of 800 or 1000 s/mm² is usually recommended for more diffusion weighting and better background suppression (25,26). The influence of different b-value combinations on ADC-based differentiation of malignant and benign tissue in patients with cervical cancer has previously been investigated, and the diagnostic accuracy was high for all, with no statistically significant differences between b-value combinations (26). A study using two b-value pairings of 0 and 600 s/mm² and of 0 and 1000 s/mm² to calculate ADCs for distinguishing cervical cancer and normal tissue, as well as different histologic types and grades, also reported equal diagnostic accuracy for the two settings (13). Second, although only one radiologist performed the image analysis, which was further evaluated by two different radiologists, intra- or inter-observer variability still remains to be a problem. Hence, further studies on inter-observer agreement or measurement repeatability might be required. Lastly, there could potentially be measurement error in the ADC calculation, because of the manual placement of the ROI. In order to minimize this error, ROIs were outlined into the solid components of the tumor as far as possible while avoiding the areas of reduced signal intensity corresponding to necrosis and ischemia within the tumor, which were similar to previous studies. However, our study is distinct in demonstrating the feasibility of ADC histogram parameters for predicting the treatment outcome for cervical cancer patients treated with definitive CRT. We also found that patients with advanced stage disease had significantly higher ADC histogram parameters (ADC75, ADC90, and ADC 95). This is important to define patients with high-risk disease that may potentially require additional systemic treatments.

In conclusion, our results reveal that ADC75, ADC90, and ADC95 of the tumor are the most significant predictors for tumor recurrence. Therefore, pre-CRT ADC histogram analysis may provide effective information for predicting tumor recurrence, and could be used to identify high-risk patients with uterine cervical cancer treated with CRT. Additionally, besides larger tumors, lower ADC75 and ADC95 values measured in pretreatment DWI are independent factors for worse OS and DFS, and patients with these features may require aggressive treatment.

Footnotes

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

References

Waggoner

. Cervical cancer. Lancet 2003; 361: 2217–2225.

Siegel

Miller

Jemal

. Cancer statistics, 2015. CA Cancer J Clin 2015; 65: 5–29.

Atahan

Onal

Ozyar

. Long-term outcome and prognostic factors in patients with cervical carcinoma: a retrospective study. Int J Gynecol Cancer 2007; 17: 833–842.

Thomas

. Improved treatment for cervical cancer–concurrent chemotherapy and radiotherapy. N Engl J Med 1999; 340: 1198–200.

Kastritis

Bamias

Efstathiou

. The outcome of advanced or recurrent non-squamous carcinoma of the uterine cervix after platinum-based combination chemotherapy. Gynecol Oncol 2005; 99: 376–382.

Thoeny

Ross

. Predicting and monitoring cancer treatment response with diffusion-weighted MRI. J Magn Reson Imaging 2010; 32: 2–16.

Van Beers

Vilgrain

. Biomarkers in abdominal imaging. Abdom Imaging 2009; 34: 663–667.

Kilickesmez

Bayramoglu

Inci

. Quantitative diffusion-weighted magnetic resonance imaging of normal and diseased uterine zones. Acta Radiol 2009; 50: 340–347.

McVeigh

Syed

Milosevic

. Diffusion-weighted MRI in cervical cancer. Eur Radiol 2008; 18: 1058–1064.

10.

Payne

Schmidt

Morgan

. Evaluation of magnetic resonance diffusion and spectroscopy measurements as predictive biomarkers in stage 1 cervical cancer. Gynecol Oncol 2010; 116: 246–252.

11.

Xue

Ren

Yang

. Histogram analysis of apparent diffusion coefficient for the assessment of local aggressiveness of cervical cancer. Arch Gynecol Obstet 2014; 290: 341–348.

12.

Nakamura

Kajitani

Joja

. The posttreatment mean apparent diffusion coefficient of primary tumor is superior to pretreatment ADCmean of primary tumor as a predictor of prognosis with cervical cancer. Cancer Med 2013; 2: 519–525.

13.

Kuang

Ren

Zhong

. The value of apparent diffusion coefficient in the assessment of cervical cancer. Eur Radiol 2013; 23: 1050–1058.

14.

Park

Kim

Park

. Assessment of early response to concurrent chemoradiotherapy in cervical cancer: Value of diffusion-weighted and dynamic contrast-enhanced MR imaging. Magn Reson Imaging 2014; 32: 993–1000.

15.

Onal

Erbay

Guler

. Treatment response evaluation using the mean apparent diffusion coefficient in cervical cancer patients treated with definitive chemoradiotherapy. J Magn Reson Imaging 2016. doi: 10.1002/jmri.25215.

16.

Punwani

. Diffusion weighted imaging of female pelvic cancers: concepts and clinical applications. Eur J Radiol 2011; 78: 21–29.

17.

Heo

Shin

Kim

. Pre-treatment diffusion-weighted MR imaging for predicting tumor recurrence in uterine cervical cancer treated with concurrent chemoradiation: value of histogram analysis of apparent diffusion coefficients. Korean J Radiol 2013; 14: 616–625.

18.

Lin

Chen

. Correlation of histogram analysis of apparent diffusion coefficient with uterine cervical pathologic finding. Am J Roentgenol 2015; 204: 1125–1131.

19.

Downey

Riches

Morgan

. Relationship between imaging biomarkers of stage I cervical cancer and poor-prognosis histologic features: quantitative histogram analysis of diffusion-weighted MR images. Am J Roentgenol 2013; 200: 314–320.

20.

Onal

Arslan

Topkan

. Comparison of conventional and CT-based planning for intracavitary brachytherapy for cervical cancer: Target volume coverage and organs at risk doses. J Exp Clin Cancer Res 2009; 28: 1–10.

21.

Schwarz

Siegel

Dehdashti

. Association of posttherapy positron emission tomography with tumor response and survival in cervical carcinoma. JAMA 2007; 298: 2289–2295.

22.

Kang

Choi

Kim

. Gliomas: Histogram analysis of apparent diffusion coefficient maps with standard- or high-b-value diffusion-weighted MR imaging–correlation with tumor grade. Radiology 2011; 261: 882–890.

23.

Chen

Zhang

Liang

. The utility of diffusion-weighted MR imaging in cervical cancer. Eur J Radiol 2010; 74: e101–106.

24.

Naganawa

Sato

Kumada

. Apparent diffusion coefficient in cervical cancer of the uterus: Comparison with the normal uterine cervix. Eur J Radiol 2005; 15: 71–78.

25.

Koh

Collins

. Diffusion-weighted MRI in the body: Applications and challenges in oncology. Am J Roentgenol 2007; 188: 1622–1635.

26.

Hoogendam

Klerkx

de Kort

. The influence of the b-value combination on apparent diffusion coefficient based differentiation between malignant and benign tissue in cervical cancer. J Magn Reson Imaging 2010; 32: 376–382.