Arch Gynecol Obstet DOI 10.1007/s00404-014-3221-9

Gynecologic Oncology

Histogram analysis of apparent diffusion coefficient for the assessment of local aggressiveness of cervical cancer Huadan Xue · Cui Ren · Jiaxin Yang · Zhaoyong Sun · Shuo Li · Zhengyu Jin · Keng Shen · Weixun Zhou 

Received: 20 July 2013 / Accepted: 14 March 2014 © Springer-Verlag Berlin Heidelberg 2014

Abstract  Purpose  To retrospectively explore the value of apparent diffusion coefficient (ADC) histogram in assessing local aggressiveness of cervical cancer. Methods  53 patients with cervical cancer, including 7 cases at stage IB1, 17 cases at stage IB2 and 29 cases at stage IIA, were subjected to preoperative MRI including diffusion-weighted imaging with b values of 0 and 800 s/ mm2. The average of mean ADC values (ADCmean), minimum ADC values (ADCmin) and the 5th to 85th percentile ADC values every 10 % (ADC5 %, ADC15 %, ADC85 %) were measured. ADC values were compared between subgroups according to pathologic subtype, histological differentiation, depth of cervical infiltration, and lymph node metastases. Results ADCmean and ADCmin for adenocarcinoma were 1,170.3 ± 97.8 × 10−6 and 748.7 ± 157.5 × 10−6 mm2 s−1, H. Xue and C. Ren contributed equally to this work. H. Xue · C. Ren · Z. Sun · S. Li · Z. Jin (*)  Department of Radiology, Peking Union Medical College Hospital, Chinese Academy of Medical Sciences and Peking Union Medical College, Shuai Fu Yuan #1, Beijing 100730, China e-mail: [email protected] J. Yang · K. Shen (*)  Department of Gynecology, Peking Union Medical College Hospital, Chinese Academy of Medical Sciences and Peking Union Medical College, Shuai Fu Yuan #1, Beijing 100730, China e-mail: [email protected] W. Zhou (*)  Department of Pathology, Peking Union Medical College Hospital, Chinese Academy of Medical Sciences and Peking Union Medical College, Shuai Fu Yuan #1, Beijing 100730, China e-mail: [email protected]

respectively, significantly higher than that of squamous cell carcinoma (SCC) (1,053.8 ± 134.3 × 10−6 and 615.6  ± 170.2 × 10−6 mm2 s−1, respectively). ADCmean and ADC5 %–ADC85 % of well or moderately tumor were significantly higher than poorly differentiated tumor, but ADCmin was not significantly different among different differentiated cervical cancer. Only ADC5 %–ADC45 % could discriminate well or moderately differentiated SCC from poorly differentiated SCC. ADC5 % for distinguishing well/ moderately from poorly differentiated cervical cancer had a largest AUC (0.83). There was no statistical difference in ADC value for different depth of cervical infiltration or lymph node metastases. Conclusions  ADC values are helpful in assessing pathologic subtype and the differentiation of cervical cancer. Keywords  DWI · ADC value · Cervical cancer · Differentiation

Introduction Cervical cancer is one of the significant threats to female health and life. 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). The most important indicators affecting the prognosis in cervical cancer are stage at diagnosis, diameter, histological subtype, grade of differentiation, depth of stromal invasion, lymphatic metastases and lymphovascular space invasion [1–3]. Accurate identification of the invasiveness of cervical cancer is important for the selection of proper treatment strategies [4]. However, there are certain difficulties to correctly estimate the tumor size by clinical examination, especially when

13



the growth is largely endocervical, parametrial or pelvic sidewall involvement, and lymph node metastases [4, 5]. Biopsy can be inaccurate compared to surgical histopathological results because of limited number of samples and heterogeneity of tumor. Non-invasive imaging techniques can provide information of entire tumor that may complement aggressive assessment by clinical staging or biopsy. Diffusion-weighted imaging (DWI) helps non-invasive characterization of biological tissues based on their water diffusion properties [6]. Water diffusion in living tissues is restricted by cellular membranes, therefore differs from the free diffusion coefficient [7]. Diffusion coefficient of water in living tissue calculated by MR examination is expressed as apparent diffusion coefficient (ADC). The ADC value is high in a tissue with little restriction to the water motion, but is low in a tissue with many obstacles. Tumors are frequently associated with architectural malformation compared to the tissue from which they originate, resulting in the shrinkage of the extracellular space and the main cause of restricted diffusion for a malignant tumor [8, 9]. ADC has been shown to correlate with histological grade of tumors, aggressiveness and clinical outcome [10–13]. However, there are few reports focusing on the relationship between ADC value and local aggressiveness of cervical cancer. Liu et al. [14] suggested that ADCmean may be helpful to identify pathologic subtype and histological differentiation of cervical cancer, but the average of ADCmean was calculated from the largest slice of the tumor and part of the pathology results from the cervical biopsy rather than radical hysterectomy. Payne et al. [15] showed that only ADC value of well/moderately differentiated cervical cancer in stage I was higher than poorly differentiated cancer, but ADC values of the tumor separated by other characteristics showed no significant difference. ADC histogram analysis can interrogate the biologic heterogeneity of tumor by classifying domains of different diffusivity, which may have prognostic and predictive implications [16, 17]. To the best of our knowledge, there has been only one study that assessed the value of ADC histogram in predicting local aggressiveness of cervical cancer, but its utility has not been obtained [18]. The aim of this study was to establish whether ADC histogram analysis is helpful in predicting local aggressiveness of cervical cancer.

Materials and methods Patient population This study was approved by the institutional review board and the requirement for informed consent in this retrospective study was waived. Sixty women with biopsy proven primary cervical cancer (mean age 43.19 years; range

13

Arch Gynecol Obstet

24–59 years) who underwent MRI examination including DWI followed by radical hysterectomy or trachelectomy between January 2008 and February 2012 were selected for this retrospective study. Surgery was performed within 2 weeks following MRI examination with no interval treatment between imaging and surgery. Two patients with cervical adenosquamous carcinoma were excluded because the sample size of this pathological type was not sufficient for statistical analysis. In addition, five patients were excluded because two staged as FIGO Ib1 could not be detected by MRI and three exhibited distorted ADC maps resulting from motion, intraluminal air or peristalsis. The final study population consisted of 53 patients (mean age 42.15 years; range 24–59 years). Imaging protocol The patients were imaged using a 1.5-T MR scanner (Signa Excite HD 1.5T, gradient strength = 40 mT/m, slew rate = 150 T/m/s; GE Healthcare, Milwaukee, WI, USA) with an eight-channel phased-array body coil. The routine pelvic exam included sagittal T2-weighted fast spin-echo (FSE) sequences (repetition time/echo time [TR/TE]  = 3,500/130 ms, slice thickness/gap = 6/1 mm, field of view (FOV) = 26 cm, matrix size = 288 × 224, NEX  = 4); axial T2-weighted FSE sequences (TR/TE  = 4,000/130 ms, slice thickness/gap = 5/0 mm, FOV = 26 cm, matrix size = 320 × 224, NEX = 4); coronal T2-weighted FSE sequences (TR/TE = 4,000/130 ms, slice thickness/gap = 6/0 mm, FOV = 30 cm, matrix size  = 288 × 224, NEX = 4); axial T1-weighted FSE sequences (TR/TE = 445/10 ms, slice thickness/ gap = 6/1 mm, FOV = 26 cm, matrix size = 288 × 192, NEX  = 2). Pelvic diffusion-weighted sequences were performed using a two- or three-station STIR–DWI–echo planar imaging (EPI) sequence with coverage from renal hilum to external aperture of vagina. Imaging parameters were as follows: TR/TE = 3,000/Minimum ms, NEX = 6, matrix  = 128 × 128, FOV = 36 cm, slice thickness/ gap = 5/0 mm. The corresponding b value to the diffusion sensitizing gradient was 0, 800 s/mm2. ADC value measurement and analysis Images were retrospectively analyzed and ADC value of each tumor was measured by two radiologists experienced in pelvic imaging, who were aware of the diagnosis of cervical cancer but fully blinded to other histological characteristics of cervical cancer. ADC maps were automatically generated from isotropic DWI by the GE AW4.2 post processing workstation with FuncTool software, and ADC histogram of every tumor slice was generated with Reformat software. ROIs were manually drawn around entire lesions

Arch Gynecol Obstet

excluding hemorrhagic, necrotic or cystic regions on each consecutive tumor containing slice of ADC maps with reference to corresponding T2-weighted images. ADC of each voxel was automatically calculated using the following formula: ADC = (lnSI0−lnSI)/b, where SI0 corresponds to signal intensity without diffusion weighting (b  = 0 s mm−2), SI is signal intensity obtained at b  = 800 s mm−2. ADCmin was the lowest ADC of one voxel within ROI, ADCn % was the point at which n % of the voxel values that form the histogram are found to the left [16]. ADCmean was the average of all the ADC values within ROI. ADC values were calculated and compared between subgroups according to their pathologic status. Histopathologic analysis Surgical specimens were assessed by the same pathologist who was blinded to MR results. Histologic specimens were fixed in 10 % buffered formalin and embedded in paraffin. Sections were stained with hematoxylin–eosin. In each case, the histological type, tumor grade, the presence of lymph node metastasis, the depth of stromal infiltration (

Histogram analysis of apparent diffusion coefficient for the assessment of local aggressiveness of cervical cancer.

To retrospectively explore the value of apparent diffusion coefficient (ADC) histogram in assessing local aggressiveness of cervical cancer...
437KB Sizes 0 Downloads 3 Views