Cervical cancer evaluated with integrated 18F-FDG PET/MR.

The current study aimed to evaluate the correlation between maximum standardized uptake value (SUVmax) and minimum apparent diffusion coefficient (ADCmin) of cervical cancer using an integrated 18F-fluorodeoxyglucose positron emission tomography/magnetic resonance (PET/MR) imaging system, and to determine the association with pathological prognostic factors. A total of 46 patients were pathologically diagnosed with cervical cancer and underwent PET/MR prior to surgery, including total hysterectomy, bilateral pelvic lymph node dissection or paraaortic lymph node dissection. The imaging biomarkers included the SUVmax and ADCmin. The pathological prognostic factors were as follows: Tumor size, histological grade, International Federation of Gynecology and Obstetrics (FIGO) stage and lymph node metastasis. Pearson's correlation analysis was used to evaluate the correlation between imaging biomarkers and the tumor size and the Mann-Whitney U test analysis was used to evaluate the association between imaging biomarkers and pathological factors. The mean SUVmax was 11.1±8.7 (range, 3.16-51.6) and the mean ADCmin was 0.76±0.15×10-3 mm2/s (range, 0.47-1.04×10-3 mm2/s). The SUVmax had a significant negative correlation with the ADCmin (r=-0.700; P<0.001). The SUVmax was significantly increased in patients with poorly differentiated tumors (P=0.001), patients with FIGO stage IIB (P=0.005) and the patients with lymph node metastasis (P=0.040). The ADCmin was significantly decreased in patients with poorly differentiated tumors (P<0.001) and patients with FIGO stage IIB (P=0.017). Statistical analysis revealed no significant correlation between the tumor size and the SUVmax (r=0.286;P=0.054), or between the tumor size and the ADCmin (r=-0.231; P=0.122). Area under the curve (AUC) analysis revealed that SUVmax had a higher diagnostic value for lymph node metastasis (AUC=0.681) and FIGO staging (AUC=0.837) compared with ADCmin, whereas ADCmin had a higher diagnostic value for the grade of pathological differentiation (AUC=0.816) compared with SUVmax (AUC=0.788). The results of the current study demonstrated that there was a significant negative correlation between SUVmax and ADCmin, which were associated with prognostic factors.

Introduction

Cervical cancer is one of the most common gynecological cancers. In China, cervical cancer is the second most common cancer in females after breast cancer (1). In 2012, ~52,7600 new cases of cervical cancer were diagnosed worldwide and 265700 patients succumbed (2). The average age of diagnosis for cervical cancer is 50 years, which is younger than that of other types of cancer, including endometrial and ovarian cancers (3). Pretreatment evaluation of patients with cervical cancer, including staging and assessment of prognostic factors, is required for predicting the prognosis and determining the optimal treatment regimen (4).

Previous studies have revealed that patient age, clinical symptoms, degree of pathological differentiation, International Federation of Gynecology and Obstetrics (FIGO) stage, status of lymph node metastasis and tumor size affect the prognosis of patients with cervical cancer (5,6). Imaging diagnosis of cervical cancer may assist clinical decision making. Positron emission tomography (PET) and magnetic resonance (MR) imaging serve important roles in the diagnosis and treatment of tumors. Diffusion-weighted imaging (DWI) is an imaging method that reflects the diffusion of water molecules through the apparent diffusion coefficient (ADC), which is negatively correlated with tumor cellularity. This method can reflect the diffusion limitation and molecular density of tumor tissue (7). It is frequently used in tumor diagnosis and prognosis evaluation (8). 18F-fluorodeoxyglucose (18F-FDG) PET reflects glucose metabolism of the tumor, as the faster the tumor cells proliferate, the higher their metabolic rate. Patients with cervical cancer manifest lesions with marked elevations in metabolic rate on PET images (9). The standardized uptake value (SUV) is used to measure the glucose uptake within a region of interest (ROI), and may be used to assess tumor aggressiveness and prognosis (10). The maximum SUV (SUVmax) may be used as an independent prognostic factor (11). SUV and ADC values are associated with a number of pathological prognostic factors and may be used to assess the prognosis of patients. The two aforementioned parameters may serve a complementary role in describing tumor characteristics and assessing the prognosis of cervical cancer. Unlike PET and MR imaging, integrated PET/MR imaging provides multiparametric functional images in a single examination, including a DWI (7). Additionally, 18F-FDG PET/MR imaging provides metabolic information. PET/MR imaging provides comprehensive information about tumor characteristics, including the location, size and extent of the lesion (12,13). Furthermore, deviations caused by physiological changes during the time interval between PET and MR imaging are avoided (7). Previous studies have revealed that integrated PET/MR provides comprehensive information and excellent image quality in the diagnosis and staging of different types of cancers, and is superior to PET alone (14–16).

Previous studies have revealed that the maximum standard uptake value (SUVmax) has a negative correlation with the minimum ADC (ADCmin) on PET/MR imaging for various types of cancer, including endometrial (11), rectal (17), cervical (18) and breast cancer (19). However, there are currently few studies on the combination of DWI and PET functional imaging for the detection of cervical cancer (13,18). The aim of the current study was to evaluate the correlation between the SUVmax and the ADCmin using an integrated PET/MR imaging system, and to establish the correlation with pathological prognostic factors.

Materials and methods

Patients

This study was approved the Ethics Committee of the Chinese PLA General Hospital and all patients signed informed consent forms. Patients were enrolled in the Chinese PLA General Hospital (Beijing, China) between April 2016 and December 2017, and follow up was conducted one time six months following surgery. All patients had undergone PET/MR imaging examinations prior to surgery, including total hysterectomy, bilateral pelvic lymph node dissection and paraaortic lymph node dissection. A total of 46 patients were pathologically confirmed to have cervical cancer (mean age, 56.5±12.1; range, 31–74). The inclusion criteria were as follows: i) Clinical suspicion of cervical cancer; and ii) histopathological diagnosis of cervical cancer following PET/MR imaging. The exclusion criteria were as follows: i) Pregnant patients; ii) patients with metal implants; iii) patients with claustrophobia, iv) patients with impaired renal function; and v) patients with poor quality PET/MR images.

Following tumor resection, tissues were fixed with 10% formaldehyde for 24 h, dehydrated in an ethanol gradient (100, 95, 85 and 75% ethanol for 5, 4, 3 and 2 min, respectively), wax-impregnated and embedded in paraffin. Serial sections (5 µm thick) were prepared and dried at 50°C for 30 min. Sections were stained with hematoxylin for 5 min and 1% eosin for 30–60 sec at room temperature. Histopathological evaluations of these sections were assessed using a light microscope (magnification, 40×). The patients received surgical treatment and the tumor size, grade, stage and the status of lymph node metastasis were recorded. The histopathologic findings were reviewed as previously described (20,21).

PET/MR examination

Whole-body PET/MR examination was performed using an integrated PET/MR machine (Biograph mMR; Siemens Healthineers). Prior to the examination, patients fasted for >6 h, and their fasting blood glucose levels were monitored. Prior to image acquisition, 18F-FDG (Sumitomo HM-20 Cyclotron) 2.22–4.44 MBq/kg was injected intravenously at rest. Images were acquired 50–60 min following injection. The acquisition time was 3 min per bed position, and the accumulated acquisition time was 40 min/person. The attenuation correction technique was used based on the Dixon sequence: Images were reconstructed with an iterative algorithm (3 iterations; 21 subsets; and a Gaussian filter with a full width at half maximum=4 mm for scatter correction) incorporating a point spread function (22). MR image C was acquired using the complete Matrix coils to cover most of the trunk (from the neck to the middle segment of the femur). The sequences included 3D volumetric interpolated breath-hold T1-weighted sequence in the horizontal axis, and the T2-weighted turbo spin-echo imaging (T2WI-TSE) sequence in the horizontal axis. DWI with b values of 50 and 800 in the horizontal axis was performed. Patients with cervical cancer were scanned with the sagittal T2WI-TSE sequence.

Image analysis

Two nuclear medicine physicians independently measured the ADCmin and SUVmax of primary tumors. For the ADC value, the large ROIs were drawn manually to cover the lesions (with abnormal enhancement) on the DWI. The lowest ADC value was defined as the ADCmin. The ROI was manually delineated by the substantial part of abnormal enhancement within the lesion on the PET image, and syngo MMWP software (version VE40A; Siemens Healthineers) automatically calculated the SUV value of the lesion. The highest SUV value of the ROI was defined as the SUVmax.

Data analysis

All data were analyzed using SPSS software (version 19.0; IBM Corp., Armonk, NY, USA). Continuous data were presented as the mean ± standard deviation. Pearson's correlation analysis was used to evaluate the correlation between imaging biomarkers and the tumor size. The Mann-Whitney U test was used to evaluate the associations between the imaging biomarkers (SUVmax and ADCmin) and pathological factors (histological grade, FIGO stage and lymph node metastasis) (18,23). The sensitivity and specificity of the imaging biomarkers for pathological factors were determined by measuring the area under the curve (AUC). P<0.05 was considered to indicate a statistically significant difference.

Results

Patient clinical data

This study enrolled a total of 46 patients with a definitive pathological diagnosis of cervical cancer. The patients had a mean age of 56.5±12.1 years (range, 31–74). Detailed clinical data of the patients are presented in Table I. A total of 40 patients had squamous carcinoma and 6 had adenocarcinoma. A total of 23 patients had poorly differentiated carcinoma, 23 had moderately differentiated carcinoma and none had well differentiated carcinoma. The mean tumor size was 10.3±9.8 mm3 (maximum size, 60 mm3). All 46 patients underwent FIGO staging, which revealed stage IB in 25 patients (moderately differentiated; Fig. 1), stage IIA in 10 patients (moderately differentiated; Fig. 2) and stage IIB in 11 patients. A total of 18 patients had cervical invasion with lymph node metastasis, including left pelvic lymph node metastasis (poorly differentiated; stage IIA; Fig. 3) and bilateral pelvic lymph node metastasis (poorly differentiated; stage IIB; Fig. 4).

Table I.

Clinical data of the 46 patients with cervical cancer in the present study.

Characteristics	Value
Age (years)	56.5±12.1 (31–74)
Tumor size (mm3)	10.3±9.8 (0.72–60)
Histology
Adenocarcinoma	6 (13)
Squamous carcinoma	40 (87)
Histology grade
Well-differentiated	0 (0)
Moderately-differentiated	23 (50)
Poorly-differentiated	23 (50)
International Federation of Gynecology and Obstetrics stage
IB	25 (54)
IIA	10 (22)
IIB	11 (24)
Lymph node metastasis
Absent	28 (61)
Present	18 (39)

The data are presented as the mean ± standard deviation (range) or number (percentage) of patients.

Figure 1.

PET/MR scans of a 55 years old patient with cervical cancer (moderately differentiated, stage IB). (A) Axial T2-weighted MR image revealed a cervical tumor (red arrow). (B) An ADC map revealed low signal intensity of the tumor (red arrow; ADCmin=0.78×10−3 mm2/s). (C) PET and (D) PET/MR hybrid images image revealed the high glucose metabolic activity of the tumor (red arrow; SUVmax=8.8). (E) Pathological images revealed moderate differentiation (magnification, 40×). PET, positron emission tomography; MR, magnetic resonance; ADC, apparent diffusion coefficient; min, minimum; SUVmax, maximum standardized uptake value.

Figure 2.

PET/MR scans of a 74 years old patient with cervical cancer (moderately differentiated, stage IIA). (A) Axial T2-weighted MR image revealed a cervical tumor (red arrow). (B) ADC map revealed the low signal intensity of the tumor (red arrow; ADCmin=0.63×10−3 mm2/s). (C) PET and (D) PET/MR hybrid images revealed the high glucose metabolic activity of the tumor (red arrow; SUVmax=15.6). (E) Pathological images revealed moderate differentiation (magnification, ×40). PET, positron emission tomography; MR, magnetic resonance; ADC, apparent diffusion coefficient; min, minimum; SUVmax, maximum standardized uptake value.

Figure 3.

PET/MR scans of a 45 years old female with cervical cancer (poorly differentiated, stage IIA). (A) Axial T2-weighted MR image revealed a primary tumor (red arrow) and a secondary tumor (hollow arrow) in the left pelvic cavity. (B) ADC map revealed the low signal intensity of the tumor (red arrow; primary tumor ADCmin=0.8×10−3 mm2/s). (C) PET and (D) PET/MR hybrid images image revealed the high glucose metabolic activity of the tumor (red arrow; primary tumor SUVmax=9). (E) Pathological images revealed shows poorly differentiation (magnification, ×40). PET, positron emission tomography; MR, magnetic resonance; ADC, apparent diffusion coefficient; min, minimum; SUVmax, maximum standardized uptake value.

Figure 4.

PET/MR scans of a 49 years old female with cervical cancer (poorly differentiated, stage IIB). (A) Axial T2-weighted MR image revealed a tumor (red arrow). (B) ADC map revealed low signal intensity of the tumor (red arrow; primary tumor ADCmin=0.67×10−3 mm2/s). (C) PET and (D) PET/MR hybrid images revealed high glucose metabolic activity of the tumor (red arrow; primary tumor SUVmax=16.3). (E) Axial T2-weighted MR image revealed the tumor in the cervix (arrow) with two metastatic lymph nodes (hollow arrow) in the right and left pelvic cavity. (F) ADC map revealed low signal intensity of the primary (red arrow) and metastatic (hollow arrow) lesions. (G) PET and (H) PET/MR hybrid images revealed high glucose metabolic activity of the tumor. (I) Pathological images revealed poor differentiation (magnification, ×40). PET, positron emission tomography; MR, magnetic resonance; ADC, apparent diffusion coefficient; min, minimum; SUVmax, maximum standardized uptake value.

Analysis of the correlations between SUVmax, ADCmin and the tumor size

Following integrated PET/MR imaging for the 46 patients, the mean SUVmax was 11.1±8.7 (range, 3.16–51.6) and the mean ADCmin was 0.76±0.15×10−3 mm2/s (range, 0.47–1.04×10−3 mm2/s). The SUVmax had a significant negative correlation with the ADCmin (r=−0.700; P=<0.001; Fig. 5A). Statistical analysis revealed no significant difference between the tumor size and the SUVmax (r=0.286; P=0.054), or between the tumor size and the ADCmin (r=−0.231; P=0.122). The results of correlation analyses are detailed in Table II.

Figure 5.

Correlation between ADCmin and SUVmax values of cervical cancer assessed by hybrid PET/MR and ROC to evaluate the diagnostic value of SUVmax and ADCmin. (A) A scatter plot revealed a significant inverse correlation between ADCmin and SUVmax (r=−0.700; P=<0.001). ROC to evaluate the diagnostic value of SUVmax and ADCmin in pathological prognostic factors in 46 patients with cervical cancer, including (B) histological grade, (C) FIGO stage and (D) lymph node metastasis. ADCmin, minimum apparent diffusion coefficient; SUVmax, maximum standardized uptake value; ROC, receiver operating characteristic curve; PET, positron emission tomography; MR, magnetic resonance; min, minimum; FIGO, International Federation of Gynecology and Obstetrics.

Table II.

Correlations between SUVmax, ADCmin and tumor size.

Parameter	SUVmax	ADCmin	Tumor size
SUVmax
r	1	−0.700	0.286
P-value		<0.001	0.054
ADCmin
r		1	−0.231
P-value			0.122
Tumor size
r			1
P-value

SUVmax, maximum standardized uptake value; ADCmin, minimum apparent diffusion coefficient; r, Pearson product-moment correlation coefficient.

Association between SUVmax/ADCmin/tumor size and tumor differentiation/FIGO stage/lymph node metastasis

The SUVmax was significantly increased in patients with poorly differentiated tumors (P=0.001), patients with lymph node metastasis (P=0.040) and patients with FIGO stage IIB (P=0.005) compared with patients with moderately differentiated tumors, patients with no lymph node metastasis and patients with early-stage (IB-IIA), respectively. The ADCmin was significantly decreased in patients with poorly differentiated tumors (P<0.001) and patients with FIGO stage IIB (P=0.017), compared with patients with moderately differentiated tumors and patients with FIGO stage IB-IIA, respectively. The ADCmin was decreased in patients with lymph node metastasis compared with those without lymph node metastasis, but the difference was not statistically significant (P=0.105). The tumor sizes were slightly larger in patients with poorly differentiated tumors (P=0.553), patients with FIGO stage IIB (P=0.193), and patients with lymph node metastasis (P=0.386), compared with patients with moderately differentiated tumors, FIGO stage IB-IIA and without lymph node metastasis, respectively, but the differences were not statistically significant (Table III).

Table III.

Correlation between the SUVmax/ADCmin/tumor size and tumor differentiation/FIGO stage/lymph node metastasis.

	Histological grade			FIGO stage			Lymph node metastasis
Parameter	Poorly differentiated (n =23)	Moderately differentiated (n =23)	P-value	(IB-IIA) (n=35)	IIB (n=11)	P-value	Absent (n=28)	Present (n=18)	P-value
SUVmax	14.9±10.6	7.3±3.6	0.001a	8.6±4.9	18.9±13.1		8.79±4.83	14.7±11.9	0.040
	(3.16–54.6)	(3.18–15.6)		(3.16–22.4)	(4.94–54.6)	0.005	(3.18–20.1)	(3.16–54.)
ADCmin	0.67±0.14	0.84±0.11	0.000a	0.79±0.14	0.66±0.15	0.017	0.78±0.14	0.72±0.17	0.105
(×10−3 mm2/s)	(0.47–0.97)	(0.62–1.04)		(0.47–1.04)	(0.48–0.97)		(0.47–1.02)	(0.48–1.04)
Tumor size	11.6±12.2	9.0±6.7	0.553	9.6±10.4	12.5±7.9	0.193	8.61±5.96	12.9±13.7	0.386
(mm3)	(2.89–60.0)	(0.72–25.0)		(0.72–60)	(2.89–25)		(3.0–25)	(0.72–60.0)

The data are presented as the means ± standard deviation (range). SUVmax, maximum standardized uptake value; ADCmin, minimum apparent diffusion coefficient; FIGO, International Federation of Gynecology and Obstetrics.

Sensitivity and specificity of SUVmax and ADCmin for histological grade, FIGO stage and lymph node metastasis are presented in Fig. 5. SUVmax exhibited a higher AUC value for predicting FIGO stage (AUC=0.837) and lymph node metastasis (AUC=0.681) compared with ADCmin for predicting FIGO stage (AUC=0.745) and lymph node metastasis (AUC=0.643). ADCmin was more useful for predicting the grade of tumor differentiation (AUC=0.816) compared with SUVmax (AUC=0.788).

Discussion

The present study demonstrated that PET/MR imaging provides valuable imaging data for patients with cervical cancer. These PET/MR results were associated with various pathological factors and may serve a role in the evaluation of the prognosis of patients with cervical cancer. The SUV represents the metabolism of glucose in tumor cells (15,23). The invasive ability of tumors may be estimated from the SUV, and this may be used to predict the prognosis to some extent (24). The ADC presents tumor cellularity and may reflect tumor invasiveness (25). The ADC may be used to indicate tumor characteristics, including tumor size, number of metastatic lymph nodes, portal vein invasion and extrapancreatic nerve plexus invasion (25). Previous studies have revealed that the SUVmax and the ADCmin may be used as tumor imaging biomarkers, and they have a significantly negative correlation in different types of cancer, including endometrial (11), rectal (17), cervical (18) and breast cancer (19). As was the case with the aforementioned studies, the current study demonstrated that the SUVmax and the ADCmin had a significantly negative correlation, which may present a biological association between tumor cellularity and glucose metabolic activity. Previous studies relied on SUV and ADC values measured separately by distinct PET and MR devices (19,23). The time interval between the two examinations may impact the analysis. By contrast, an integrated PET/MR system provides the two types of data at the same time (26). The current study revealed that the SUVmax and the ADCmin had no significant correlation with the tumor size, as did a previous study by Kidd et al (27). Notably, a P-value close to 0.05 was obtained (P=0.054), and this warrants further investigation by expanding the sample size in future studies, as it may be used for guiding clinical diagnosis.

The grade of pathological differentiation and the FIGO stage are key factors affecting prognosis of patients with cervical cancer (28). The SUVmax and ADCmin are associated with pathological prognostic factors in various types of cancer (11,17,25). Previous studies revealed that tumor FDG uptake was associated with the grade of pathological differentiation: Poorly differentiated tumors had a higher SUVmax than moderately and well differentiated tumors (9,19,23,27). A previous study revealed that patients with poorly differentiated tumors had a decreased ADC compared with patients with well differentiated tumors (29). In addition, a previous study revealed that the SUV was positively correlated with the FIGO stage (4). The current study revealed a significantly increase in the SUVmax and a significant decrease in the ADC value in patients with poorly differentiated tumors compared with those with moderately differentiated tumors, as well as significant differences in the SUVmax and ADCmin values between patients with early cervical cancer (stage IB-IIA) and patients with mid-stage cervical cancer (stage IIB). The SUV in patients with stage (IIB-IIA) was decreased compared with patients with stage IIB. These results may be due to an increase in glucose uptake caused by the rapid growth and active proliferation of tumors, resulting in an increased FDG uptake in patients with poorly differentiated tumors.

The presence or absence of lymph node metastasis is an important prognostic factor in patients with early cervical cancer (9). Previous studies revealed SUV is significantly associated with lymph node metastasis (6,30). A previous study revealed that patients with cervical cancer with lymph node metastasis, had an increased SUV in their primary lesions compared with patients without lymph node metastasis (31). The current study demonstrated that a higher SUV was positively associated with lymph node metastasis, which indicated that high FDG uptake may predict local invasion of tumor cells. There was no significant difference in ADCmin between patients with lymph node metastasis and those without lymph node metastasis in the current study, contrary to observations made by Chen et al (32), who demonstrated that patients with lymph node metastasis had a significantly decreased ADCmin compared with patients without lymph node metastasis. This difference may be attributed to a smaller sample size.

Surgery is the preferred treatment for patients with early cervical cancer (4,13). By analyzing imaging markers to evaluate prognostic factors, surgical approaches may be optimized for individual patients (31,33). The current study demonstrated that PET had a higher diagnostic value for lymph node metastasis and FIGO staging than MR imaging alone, whereas MR imaging had a higher diagnostics value for the grade of pathological differentiation compared with PET. Thus, the comprehensive image data provided by an integrated PET/MR imaging system may have greater clinical significance than the data provided by MR imaging or PET alone.

There are two limitations for the current retrospective study. Firstly, no analysis was performed for the cervical cancer based on the pathological types due to the small sample size and future studies with a larger sample are required. Secondly, due to short follow-up time subsequent to the examination the postoperative survival of the patients was not included in the current study. The prognostic significance of the SUV and ADC were therefore not demonstrated.

In conclusion, the results of the current study revealed that there was a significant negative correlation between SUVmax and ADCmin. These two imaging biomarkers were associated with pathological prognostic factors. Compared with PET or MR imaging alone, integrated PET/MR imaging may provide higher value data for the diagnosis of cervical cancer.