Metabolic tumour volume and total lesion glycolysis, measured using preoperative 18F–FDG PET/CT, predict the recurrence of endometrial cancer




To investigate the prognostic value of metabolic tumour volume (MTV) and total lesion glycolysis (TLG), measured by preoperative positron emission tomography and computerised tomography (PET/CT), in women with endometrial cancer.


Retrospective cohort study.


A tertiary referral centre.


Women with endometrial cancer who underwent preoperative 18F-FDG PET/CT in the period 2004–2009.


Clinicopathological data for 84 women with endometrial cancer were reviewed from medical records. Cox proportional hazards modelling identified recurrence predictors. The receiver operating characteristic (ROC) curve was used to determine the cut-off value for predicting recurrence.

Main outcome measure

Disease-free survival (DFS).


The number of patients with International Federation of Gynecology and Obstetrics (FIGO) stages were: I (58); II (11); III (13); and IV (2). The median DFS was 48 (1–85) months. By univariate analysis, DFS was significantly associated with FIGO stage, histology, peritoneal cytology, myometrial invasion, nodal metastasis, serum CA-125, MTV, and TLG. Using multivariate analysis, the MTV (= 0.010; hazard ratio, HR = 1.010; 95% confidence interval, 95% CI = 1.002–1.018) and TLG (= 0.024; HR = 1.001; 95% CI = 1.000–1.002) were associated with DFS. The area under the ROC curve was 0.679 (95% CI = 0.505–0.836) after discriminating for recurrence using an MTV cut-off value of 17.15 ml. Regarding TLG, the cut-off value was 56.43 g and the area under the ROC plot was 0.661 (95% CI = 0.501–0.827). Kaplan–Meier survival graphs demonstrated a significant difference in DFS between groups categorised using the cut-off values for MTV and TLG (< 0.022 for MTV and < 0.047 for TLG, by log-rank test).


Preoperative MTV and TLG could be independent prognostic factors predicting the recurrence of endometrial cancer.


Endometrial cancer (EC) is the most common gynaecological malignancy in Western countries, with more than 40 000 cases diagnosed annually in the USA.[1] The number of patients with EC is rapidly increasing in Korea, where it is third most common gynaecological malignancy, with an estimated 1616 new cases in 2010.[2]

Although the majority of EC patients present with early-stage disease and a favourable prognosis, some patients have aggressive disease with poor prognostic factors. Several prognostic factors have been identified for EC, including International Federation of Gynecology and Obstetrics (FIGO) stage, histology, grade, myometrial invasion, tumour size, peritoneal cytology, lymphovascular space invasion (LVSI), and lymph node metastasis (LNM).[3, 4] Imaging modalities, including computed tomography (CT) and magnetic resonance imaging (MRI), and the use of serum tumour markers, such as cancer antigen 125 (CA-125), have been studied for potential roles as preoperative assessment tools. Although these preoperative evaluations can detect LNM, myometrial invasion, and extrauterine disease, their roles in predicting prognosis are controversial.[5-9]

Metabolic tumour volume (MTV) and total lesion glycolysis (TLG), measured by18F-fluorodeoxyglucose (18F-FDG) positron emission tomography/computed tomography (PET/CT), are functional indices describing metabolic tumour burden,[10] whereas standardised uptake values (SUVs) only describe 18F-FDG uptake within a manually drawn volume of interest (e.g. SUVmean, SUVpeak) or single pixel (SUVmax). Accordingly, MTV and TLG can reflect the tumour burden more accurately than SUV. These parameters were recently reported to have prognostic value in several cancers, including lung, cervix, and ovary; however, there are limited data regarding EC.[11-15]

We speculated that MTV and TLG, determined using preoperative PET/CT, could be prognostic factors predicting recurrence in EC patients. This study was designed to identify the prognostic value of these parameters for predicting recurrence and disease-free survival (DFS) in EC patients who underwent primary surgery.



This is a retrospective study using a database of patients diagnosed with primary EC in Asan Medical Centre between October 2004 and February 2009, which was approved by our institutional review board (S2012-1754-0001). Initial PET/CT has been recommended for all EC patients for their preoperative evaluation since the introduction of PET/CT at our medical centre. Therefore, PET/CT was performed consecutively in these patients, except for patients who refused this recommendation.

The eligibility criteria were as follows: age >18 and <80 years; presence of pathologically confirmed primary EC; and PET/CT performed at our institution within 4 weeks prior to surgery. Patients who did not undergo primary surgery or follow-up at our institution, received neoadjuvant chemotherapy or preoperative radiotherapy, had a short follow-up (<3 months), and/or a history of another malignancy or underlying disease potentially affecting survival were excluded.

PET/CT scanning procedure

All patients were instructed to avoid strenuous exercise for 24 hours before the 18F-FDG PET/CT study was performed in order to minimise the uptake of the radiotracer into their muscles. They were also instructed to fast for at least 6 hours prior to the injection of 18F-FDG, which was produced in our radiopharmacy using standard synthetic techniques. Furosemide (40 mg tablet) and duspatalin (135 mg tablet) were orally administered just before venous blood glucose measurement. Venous blood glucose levels were maintained at <140 mg/dl. All patients were injected with 0.2 mCi/kg of 18F-FDG and allowed to rest in a sitting or supine position for approximately 60 minutes prior to scanning. The patients were then positioned in the scanner with their arms above their heads. PET/CT scans from the base of the skull to the mid-thigh were performed using a Discovery STE (GE Healthcare, Waukesha, WI, USA), Biograph Truepoint 16 (Siemens/CTI, Knoxville, TN, USA), or Biograph Truepoint 40 (Siemens/CTI) scanner. These scanners obtained combination multislice CT and PET tomographs. The CT data were used for attenuation correction. A total of five or six bed positions for 2–3 minutes per bed position were acquired for emission scanning (3 minutes/position for Discovery STE and Biograph Truepoint 16; 2 minutes/position for Biograph Truepoint 40). All scans were reconstructed using an ordered-subsets expectation maximisation algorithm (20 subsets and two iterations for Discovery STE; 16 subsets and two iterations for Biograph Truepoint 16; 21 subsets and three iterations for Biograph Truepoint 40). Calibration of each scanner against the dose calibrators and well counters was routinely performed. The measured SUV of the phantom was within the acceptable range of 90–110%. The mean SUV of the liver was also calculated by drawing a three-dimensional region of interest, 3 cm in diameter, within the normal inferior right lobe.

PET/CT imaging interpretation

In relation to the present study, all PET/CT images were reviewed by two experienced nuclear medicine doctors who had no knowledge of the patient's clinical information. Artifactual and physiological soft tissue accumulation of 18F-FDG was taken into account during image interpretation. The maximum standardised uptake value (SUVmax) served as a semiquantitative measurement of glucose metabolism. SUV was calculated using the following formula: SUV = [concentration of radioactivity in the volume of interest, VOI (MBq/ml)] × [total body weight/injected radioactivity (g/MBq)]. SUVmax was the highest SUV of a voxel within a specific VOI. MTV was measured from attenuation-corrected PET/CT images using an SUV-based automated contouring program (Advantage Workstation, GE Healthcare, or TrueD, Siemens). Initially, PET/CT data were transferred into the workstation in Digital Imaging and Communications in Medicine (DICOM) format, and intensity values were automatically converted to SUVs. Nuclear medicine doctors interactively selected each hypermetabolic lesion by clicking on its projection using a graphical user interface. Then, the boundaries were drawn large enough to incorporate each target lesion in the axial, coronal, and sagittal PET/CT images. In case of discordance, nuclear medicine doctors had to agree on a common result. As previously reported, we used a fixed SUV of 2.5 as the threshold for determining the contouring margins around the tumour.[16-18] SUVmean (the mean SUV within the VOI) and MTV of the primary tumour inside the boundaries were automatically calculated. TLG was calculated as SUVmean × MTV of the target lesion. For the ten cases demonstrating no definite hypermetabolic activity above the threshold SUV in the uterus, we considered both MTV and TLG of the primary tumour to be zero.

Surgical staging and pathological evaluation

The surgical staging procedures for EC in our centre have been described previously.[19] All patients underwent at least a total abdominal hysterectomy (TAH) or laparoscopic-assisted vaginal hysterectomy (LAVH), with or without bilateral salpingo-oophorectomy (BSO). In cases with invasion equal to or greater than half of the myometrium, large tumour size (>2 cm in diameter), grade 2–3, FIGO stage II–IV, or non-endometrioid histology, a comprehensive surgical staging procedure, including pelvic and paraaortic lymph node dissection, was performed. Each primary tumour and lymph node was sliced and stained with haematoxylin and eosin, and was microscopically examined by a pathologist. The number of lymph nodes retrieved from each area and the presence of metastasis were recorded. Adjuvant treatment was selected based on the FIGO stage, histology, grade, and doctor's discretion.

Follow-up evaluation

After completing primary treatment, all patients were evaluated by a gynaecological oncologist at 1 month, at 3-month intervals for 2 years, and every 6 months thereafter. A diagnosis of recurrence was based on either tissue biopsy or the demonstration of radiological evidence of progression.

Variables and statistical analysis

Disease-free survival (DFS) was calculated in months from the date of the patient's surgery to either the date of recurrence or the date the patient was censored. Patients were censored from the survival analysis if they were alive at last contact or had died without disease recurrence. To analyse the prognostic variables associated with DFS, we assessed the following factors: age; FIGO stage; histology; grade; peritoneal cytology; tumour size; myometrial invasion depth; lymphovascular space invasion (LVSI); lymph node metastasis (LNM); preoperative serum CA-125; SUVmax, SUVmean, MTV, and TLG of the primary tumour. Univariate and multivariate analyses of potential prognostic factors for DFS were performed using Cox proportional hazards regression. Multicollinearity between variables was evaluated using the Pearson correlation coefficient and variance inflation factor (VIF). Age, tumour size, CA-125, SUVmax, SUVmean, MTV, and TLG were modelled as continuous variables for the statistical analysis. Peritoneal cytology, LVSI, and LNM were considered as dichotomous variables. Similarly, the FIGO stage, histology, grade, and myometrial invasion depth were categorised as follows: stages I and II versus stages III and IV; endometrioid versus non-endometrioid type; grade 1 versus grades 2–3; and <1/2 versus ≥1/2. Receiver operating characteristic (ROC) curve analysis was performed to determine the cut-off values of MTV and TLG for predicting recurrence using Youden's index.[20] The sensitivity and specificity of each MTV and TLG in predicting the gold standard were calculated, and the cut-off point showing the maximal sum of sensitivity and specificity was determined to be the significant cut-off. Survival curves were estimated using the Kaplan–Meier method, and differences in survival were assessed using the log-rank test. Mean values between groups were compared using the Student's t-test or the Mann–Whitney U-test. P < 0.05 according to two-sided tests indicated significant difference. All analyses were performed using spss 13.0 (SPSS, Chicago, IL, USA).


Patient characteristics

During the study period, 277 patients with EC received primary surgery at our institution. Of these, 87 underwent preoperative PET/CT scans. A total of 84 patients met the inclusion criteria (Figure 1). The characteristics of the enrolled patients are summarised in Table 1. The median age of the patients was 51 years (range: 24–76 years), and the median body mass index was 24.5 kg/m2 (range: 17.3–40.2 kg/m2). Fifty-eight patients were classified as FIGO stage I, and 63 patients had endometrioid histology. Forty-four patients received adjuvant treatment after surgery: radiotherapy (n = 17); chemotherapy (n = 21); or concurrent chemoradiation (n = 6).

Table 1. Clinicopathological patient characteristics
CharacteristicsPatients (n = 84)
  1. BSO, bilateral salpingo-oophorectomy; LAVH, laparoscopic-assisted vaginal hysterectomy; LRH, laparoscopic radical hysterectomy; PALND paraaortic lymph node dissection; PLND, pelvic lymph node dissection; RH, radical hysterectomy.

Age, yearsMedian (range)51 (24–76)
BMI, kg/m2Median (range)24.5 (17.3–40.2)
FIGO stage, n (%)IA45 (53.5)
IB13 (15.5)
II11 (13.1)
IIIA2 (2.4)
IIIC11 (13.1)
IVA2 (2.4)
Histological type, n (%)Endometrioid63 (75.0)
Papillary serous10 (11.9)
Carcinosarcoma9 (10.7)
Clear cell1 (1.2)
Adenosquamous1 (1.2)
Grade, n (%)131 (36.9)
225 (29.8)
328 (33.3)
Surgical approach, n (%)Laparotomy31 (36.9)
Laparoscopy53 (63.1)
Procedures performed, n (%)TAH22 (26.2)
LAVH50 (59.5)
RH9 (10.7)
LRH3 (3.6)
BSO77 (91.7)
PLND79 (94.0)
PALND51 (60.7)
Peritoneal cytology, n (%)Negative72 (85.7)
Positive6 (7.1)
Tumour diameter, cm Depth of myometrial invasion, n (%)Median (range)3.5 (0–10.5)
<1/256 (66.7)
≥1/228 (33.3)
LVSI, n (%)Absent63 (75)
Present21 (25)
Lymph node metastasisAbsent66 (83.5)
Present13 (16.5)
Preoperative serum CA-125, U/mlMedian (range)12.2 (2.6–1950.0)
SUVmaxMedian (range)8.1 (1.30–47.9)
SUVmeanMedian (range)4.6 (2.7–15.7)
MTV, mlMedian (range)13.1 (0–379.2)
TLG, gMedian (range)57.6 (0–2367.0)
Adjuvant treatment, n (%)No40 (47.6)
Radiotherapy only17 (20.2)
Chemotherapy only21 (25.0)
Concurrent chemoradiation6 (7.1)
Recurrence, n (%) 12 (14.3)
Follow-up, monthsMedian (range)49 (3–85)
Figure 1.

Flow chart illustrating the inclusion of patients.

18F-FDG PET/CT and pathological variables

Table 2 shows the mean values of SUVmax, MTV, and TLG of the primary tumours according to pathological variables. The mean MTV of the primary tumour was significantly higher in patients with advanced FIGO stage (= 0.011), grade 2–3 (= 0.045), tumour size ≥ 4 cm (< 0.001), depth of myometrial invasion ≥ 1/2 (< 0.001), LVSI (P = 0.005), and LNM (= 0.034). The mean TLG of the primary tumour was significantly higher in patients with advanced FIGO stage (= 0.020), tumour size ≥ 4 cm (< 0.001), depth of myometrial invasion ≥ 1/2 (< 0.001), and LVSI (P = 0.007). Meanwhile, SUVmax was significantly higher only in patients with tumour size ≥ 4 cm (< 0.001) and depth of myometrial invasion ≥ 1/2 (< 0.001).

Table 2. Comparison of SUVmax, MTV, and TLG values of primary tumours according to pathological variables
CharacteristicsPatients (n = 84)SUVmaxMTVTLG
Mean ± SD P Mean ± SD P Mean ± SD P
  1. E, endometrioid histology; Non-E, non-endometrioid histology (papillary serous, carcinosarcoma, clear cell, adenosquamous).

FIGO stageI, II699.41 ± 8.030.28731.72 ± 55.710.011203.27 ± 396.260.020
III, IV1510.55 ± 6.6265.71 ± 95.91349.253 ± 559.34
Histological typeE639.77 ± 8.170.97933.28 ± 62.860.320219.69 ± 464.820.374
Non-E219.20 ± 6.6551.47 ± 72.62259.86 ± 316.28
Grade1319.25 ± 9.770.75822.18 ± 29.660.045141.93 ± 234.490.099
2, 3539.82 ± 6.5646.59 ± 77.46278.15 ± 501.95
Peritoneal cytologyNegative729.49 ± 7.860.60534.04 ± 55.900.429212.04 ± 394.240.396
Positive610.77 ± 7.7689.13 ± 147.45492.88 ± 850.82
Tumour size<4 cm476.56 ± 6.12<0.0015.49 ± 7.78<0.00128.11 ± 49.56<0.001
≥4 cm3713.26 ± 8.0277.40 ± 81.67475.49 ± 548.15
Depth of myometrial invasion<1/2567.81 ± 7.85<0.00115.61 ± 25.00<0.00195.03 ± 183.77<0.001
≥1/22813.05 ± 6.4381.01 ± 93.45490.235 ± 618.22
LVSIAbsent639.14 ± 8.250.06232.59 ± 59.000.005206.28 ± 415.130.007
Present2110.99 ± 6.1553.48 ± 81.26298.80 ± 475.17
Lymph node metastasisAbsent669.82 ± 8.010.76733.77 ± 56.550.034215.37 ± 402.500.064
Present139.48 ± 5.7670.23 ± 102.52367.49 ± 599.67

Survival analysis

After a median follow-up period of 49 months (range: 3–85 months), 12 (14.3%) patients had demonstrated recurrence and nine (10.7%) patients had died of cancer. The median DFS was 48 months (range: 1–85 months). According to the univariate analysis, DFS was significantly associated with FIGO stage (< 0.001), histology (= 0.002), peritoneal cytology (< 0.001), myometrial invasion depth (= 0.012), LVSI (= 0.029), LNM (< 0.001), CA-125 (= 0.003), MTV (P = 0.001), and TLG (= 0.036; Table 3). However, SUVmax and SUVmean values of the primary tumour did not influence DFS according our univariate analysis. Although tumour size demonstrated borderline significance for DFS (= 0.052), we included it in the multivariate analysis because it is a well-known prognostic factor. Before multivariate analysis, the multicollinearity between the variables was evaluated. Strong correlation between MTV and TLG (= 0.911 and 0.830, respectively) was detected and multicollinearity was suspected based on the calculation of VIF for MTV and TLG (VIF = 9.446 and 8.227, respectively). Therefore, MTV and TLG were incorporated into separate models in the multivariate analysis. In model 1, DFS was significantly influenced by histology (= 0.033), peritoneal cytology (= 0.007), CA-125 (= 0.008), and MTV (= 0.010; Table 4). In model 2, DFS was significantly influenced by histology (= 0.044), peritoneal cytology (= 0.003), CA-125 (= 0.005), and TLG (= 0.024; Table 5).

Table 3. Univariate analysis of the prognostic factors associated with disease-free survival
VariablesHR (95% CI) P
  1. E, endometrioid histology; Non-E, non-endometrioid histology (papillary serous, carcinosarcoma, clear cell, adenosquamous).

  2. a

    Considered a continuous variable.

Agea 1.049 (0.981–1.123)0.163
Surgical approachLaparotomy1
Laparoscopy0.369 (0.117–1.165)0.089
FIGO stageI, II1
III, IV9.394 (2.962–29.794)<0.001
Histological typeE1
Non-E6.989 (2.102–23.239)0.002
2, 346.585 (0.392–5542.121)0.115
Peritoneal cytologyNegative1
Positive12.039 (3.443–42.103)<0.001
Tumour sizea 1.230 (0.999–1.515)0.052
Depth of myometrial invasion<1/21
≥1/24.676 (1.406–15.548)0.012
Positive3.535 (1.139–10.978)0.029
Lymph node metastasisNegative1
Positive9.049 (2.743–29.851)<0.001
Preoperative serum CA-125a 1.002 (1.001–1.003)0.003
SUVmaxa 0.990 (0.916–1.070)0.805
SUVmeana 0.923 (0.670–1.273)0.626
MTVa 1.009 (1.004–1.015)0.001
TLGa 1.001 (1.000–1.002)0.036
Table 4. Multivariate analysis of the prognostic factors associated with disease-free survival (model 1a)
VariablesHR (95% CI) P
  1. E, endometrioid histology; Non-E, non-endometrioid histology (papillary serous, carcinosarcoma, clear cell, adenosquamous).

  2. a

    TLG was not incorporated into model 1 of the multivariate analysis because of multicollinearity between MTV and TLG (r = 0.911 and 0.830, respectively).

  3. b

    Considered a continuous variable.

FIGO stageI, II10.736
III, IV0.505 (0.009–26.837)
Histological typeE10.033
Non-E6.194 (1.154–33.228)
Peritoneal cytologyNegative10.007
Positive8.714 (1.788–42.458)
Tumour sizeb 0.859 (0.513–1.439)0.565
Depth of myometrial invasion<1/2 
≥1/21.709 (0.261–11.197)0.576
Positive1.208 (0.176–8.306)
Lymph node metastasisNegative10.387
Positive2.085 (0.394–11.039)
Preoperative serum CA-125b 1.002 (1.000–1.004)0.008
MTVb 1.010 (1.002–1.018)0.010
Table 5. Multivariate analysis of the prognostic factors associated with disease-free survival (model 2a)
VariablesHR (95% CI) P
  1. E, endometrioid histology; Non-E, non-endometrioid histology (papillary serous, carcinosarcoma, clear cell, adenosquamous).

  2. a

    MTV was not incorporated in model 2 of the multivariate analysis because of multicollinearity between MTV and TLG (r = 0.911 and 0.830, respectively).

  3. b

    Considered a continuous variable.

FIGO stageI, II10.749
III, IV0.515 (0.009–29.720)
Histological typeE10.044
Non-E5.969 (1.049–33.954)
Peritoneal cytologyNegative10.003
Positive9.944 (2.140–46.202)
Tumour sizeb 0.985 (0.606–1.601)0.952
Depth of myometrial invasion<1/2 
≥1/21.228 (0.213–7.096)0.818
Positive1.218 (0.183–8.086)0.838
Lymph node metastasisNegative1
Positive2.609 (0.509–13.357)0.250
Preoperative serum CA-125b 1.002 (1.001–1.003)0.005
TLGb 1.001 (1.000–1.002)0.024

Figure 2 shows the ROC curve analyses performed to determine the cut-off values for MTV and TLG that predict recurrence. The area under the ROC plot for discriminating recurrence, using an MTV cut-off value of 17.15 ml, was 0.679 (95% confidence interval, 95% CI = 0.505–0.836; sensitivity, 75.0%; specificity, 58.6%; false negative, three of 45; false positive, 29 of 39; negative predictive value, 93.2%; positive predictive value, 23.7%; Figure 2A). The area under the ROC plot for discriminating recurrence using the TLG cut-off value of 56.43 g was 0.661 (95% CI = 0.501–0.827; sensitivity, 75.0%; specificity, 52.9%; false negative, three of 42; false positive, 33 of 42; negative predictive value, 92.7%; positive predictive value, 20.9%; Figure 2B). There were three, false- negatively identified patients who had a relapse with an MTV cut-off value of <17.15 ml. They also had a TLG cut-off value of <56.43 g. All of these patients had papillary serous histology, which is well known to be a poor prognostic factor in EC. Kaplan–Meier curves for DFS in patients with MTV < 17.15 ml and ≥17.15 ml, and in patients with TLG < 56.43 g and ≥56.43 g, are presented in Figure 3A and B, respectively. There were statistically significant survival differences between the two groups that were categorised by MTV and TLG (< 0.022 for MTV and < 0.047 for TLG, according to the log-rank test).

Figure 2.

Receiver operating characteristics (ROC) curve analyses for determining the cut-off values of MTV and TLG for predicting recurrence. (A) The area under the ROC plot for discriminating recurrence using an MTV cut-off value of 17.15 ml was 0.679. (B) In the case of TLG, the cut-off value was 56.43 g and the area was 0.661.

Figure 3.

Kaplan–Meier plots of the disease-free survival rates of 84 patients with endometrial cancer who underwent preoperative PET/CT and were classified according to MTV (A; MTV < 17.15 ml versus MTV ≥ 17.15 ml) and TLG (B; TLG < 56.43 g versus TLG ≥ 56.43 g), respectively. There were statistically significant survival differences between the two groups regarding MTV and TLG (< 0.022 for MTV and < 0.047 for TLG, according to the log-rank test).


Main findings

We investigated the prognostic values of MTV and TLG measured by preoperative 18F-FDG PET/CT in EC. Both correlate with known prognostic variables and are significant prognostic factors for DFS in EC patients with primary surgery.


This is the first report of the clinical value of MTV and TLG measured by preoperative PET/CT in stage I–IV EC. This study benefits from presenting the outcomes of a single institution, wherein PET/CT scanning protocols and management policies have remained unchanged. To measure MTV and TLG, all PET/CT images were reviewed by two experienced nuclear medicine doctors who had no knowledge of the patients' clinical information. Moreover, we used the commercial automated contouring software algorithm available. Therefore, the MTV and TLG measurement methods of this study are reliable, easily adopted, and applied to clinical practice.


Previous studies reported the prognostic value of preoperative PET/CT in EC; however, these studies only evaluated SUVmax. A study of 106 EC patients by Nakamura et al.[21] found a significant correlation between primary tumour SUVmax and well-known prognostic factors. Specifically, SUVmax was an independent prognostic factor of overall survival. Similarly, Kitajima et al.[22] analysed 57 EC patients who underwent preoperative PET/CT. SUVmax was significantly associated with recurrence in multivariate analysis. However, our study demonstrated that SUVmax does not influence DFS, whereas MTV and TLG are significant prognostic factors of DFS. This may be because of the small number of recurrences and different tumour characteristics in our study population.

As SUVmax only represents the single greatest point of metabolic activity within the tumour, it cannot evaluate the entire metabolic tumour burden. Meanwhile, MTV and TLG evaluate metabolic activity throughout the tumour volume. Therefore, these parameters could reflect tumour biology, prognosis, and treatment response more precisely than SUVmax. Liu et al.[23] retrospectively analysed 15 patients with stage-IVB EC who underwent PET/CT before treatment and found that total-body MTV and TLG are significant prognostic overall survival factors. The study was limited by its small sample size and its focus on stage IVB.

As MTV and TLG are functional indices that consider tumour volume, they intuitively represent tumour size. In our study, these PET/CT parameters were prognostic, although tumour size was not prognostic according to multivariate analysis. This finding suggests that tumour size does not correctly represent tumour biology, and that both MTV and TLG represent not only tumour volume but also biological behaviour. Several reports suggest that 18F-FDG uptake in cancer cells is determined by several biological variables: microvessels to provide glucose; glucose transporter-1 to transport 18F-FDG into cells; number of tumour cells; tumour cell proliferation; and angiogenesis.[24-26] Accordingly, MTV and TLG would correlate with cellular proliferation and tumour aggressiveness.

In our study, SUVmax was only significantly associated with tumour size and myometrial invasion; however, MTV and TLG were significantly associated with FIGO stage, tumour size, myometrial invasion, and LVSI. Therefore, we suggest that MTV and TLG are superior to SUVmax as prognostic variables. In addition, both the group with MTV ≥ 17.15 ml and the group with TLG ≥ 56.43 g were significantly associated with LNM (data not shown). After using several predictors of LNM, including these PET/CT parameters, a prediction model for LNM such as the nomogram could be developed.

A problem associated with MTV and TLG measurement in endometrial tumours may be tracer activity in the urine, bladder, and ureters. In most patients, hypermetabolic portions of the uterine endometrium are clearly separated from urine activity in the bladder, ureters, and urethra by the use of combined transaxial, sagittal, and coronal CT images, as PET has higher resolution than the myometrial wall thickness. In the case of an indistinct margin, we attempted to not include physiological urine activity in the VOI referring to all available images, including MRI and contrast-enhanced abdominopelvic CT. There is currently no consensus regarding the optimal technique to measure MTV.[27] Discrepancies and variabilities in MTV in certain solid tumours are reported with various measurement techniques, including EC; however, MTV is a prognostic indicator regardless of the technique.[27-29] We introduced a threshold method using the arbitrary value of 2.5 because it is a commonly used standard for malignant lesions.[16-18] Although the chosen method for measuring MTV in our study might not be optimal, our results demonstrating that MTV and TLG have prognostic value in EC remain valid. We calculated cut-off values for MTV and TLG to discriminate recurrence; however, these values cannot be applied in clinical practice until external validation is performed and optimal methods for measuring MTV are established.

It was not possible to exactly differentiate tumour tissue from physiological endometrial uptake or tumour-associated endometrial change. Physiological endometrial uptake in premenopausal women can reach an SUV of 8.2.[30] Tumor-associated endometrial hypermetabolism has been reported in the endometrium near uterine cervical cancer, even in the absence of tumour invasion.[30] Therefore, those false positives may have caused MTV or TLG to be overestimated. As we used a cut-off SUV of 2.5, most of those changes are probably excluded from the VOI. Although our method can possibly ignore tumours with SUV < 2.5, we considered that tumours with low 18F-FDG uptake have little impact on prognosis.[22] Therefore, we considered MTV and TLG in ten patients with SUVmax < 2.5 to be zero, and these patients were included in the analysis. MTV and TLG remained independent prognostic factors even after these ten patients were excluded. Among these ten patients, seven had endometrioid histology and three had non-endometrioid histology (papillary serous in two and carcinosarcoma in one). Five patients had grade-1 tumours, three had grade 2, and two had grade 3. These findings show a similar distribution of histology and grade to that of the total patients enrolled. Therefore, 18F-FDG PET/CT could be used not only when SUVmax > 2.5, but also when SUVmax < 2.5.


Our study has limitations inherent to retrospective studies. First, preoperative PET/CT was not performed in every case. Thus, the study population may not represent all patients with EC and a selection bias may exist. We further compared the clinicopathological variables of patient groups based on preoperative PET/CT: the two groups did not differ significantly in clinicopathological variables (Table S1). Additionally, this study was conducted in a single institution with a small number of patients, and only 12 patients demonstrated recurrence. This may explain why well-known prognostic factors were significantly associated with DFS by univariate analysis, but not by multivariate analysis. Furthermore, we could not perform the overall survival analysis because there were only nine cases of disease-related death (Table S2); however, when we constructed Kaplan–Meier plots of the overall survival rates of the groups classified by MTV and TLG, there were survival differences, although these were of no statistical significance (Figure S1). Prospective studies with a larger number of patients and longer follow-up periods are required for confirmation. Finally, compared with that seen in Western countries, there was a lower BMI in our study population. Results might differ in an obese population (Table S3). Validation studies are necessary for targeting patients with a high BMI.


Both MTV and TLG measured using preoperative 18F-FDG PET/CT are significant prognostic factors predicting DFS in EC. After external validation, these parameters may help in patient stratification to determine who needs frequent follow-up or tailored adjunctive treatment, or to enroll patients into future clinical trials.

Disclosure of interests

The authors have no conflicts of interest or financial ties to disclose.

Contribution to authorship

Study concept and design: S.S. and D.K. Acquisition of data: S.S., D.K., D.L., S.L., J.P., J.L., J.K., Y.K., Y.K., and J.N. Analysis and interpretation of data: S.S., D.K., D.L., J.P., J.L., and J.N. Drafting of the article: S.S., D.K., D.L., S.L., J.P., J.L., J.K., Y.K., Y.K., and J.N. Study supervision: J.L. and J.N.

Details of ethics approval

This study was conducted according to the institutional and ethical rules concerning research, and was exempt from patient informed consent. No formal ethics approval was required.


No funding source.



Mini commentary on ‘Metabolic tumour volume and total lesion glycolysis, measured using preoperative 18F-FDG PET/CT, predicts the recurrence of endometrial cancer’

  • S Ghaem-Maghami

  • Imperial College, London, UK

Endometrial cancer often presents early and, in the majority of cases, carries a good prognosis in terms of disease-free and overall survival. Despite presenting at an early stage, for a minority of women the disease is aggressive with a poor long-term outcome (Morneau et al. Gynecol Oncol 2013;131:231–40). In order to identify these women early so that treatment can be tailored appropriately, pathological factors such as tumour type, stage, and grade have been explored, but have limitations as predictors.

To address this, Shim et al. have used novel metabolic PET/CT measurements to predict prognosis, which they have correlated with pathological features of endometrial cancer. Their cohort appears to be typical, with the majority of patients presenting with stage-I disease. In this elegant study they demonstrate a significant correlation between metabolic tumour volume and total lesion glycolysis with disease-free survival. One or both of these measures was correlated with higher disease stage and grade, tumour size, depth of invasion, the presence of lymphovascular space invasion (LVSI), and lymph node metastasis. Both of these measurements showed negative predictive values of around 93% for recurrence, which may have important implications for the stratification of patients to adjuvant therapy.

This study illustrates that PET imaging using 18F-FDG, a radioactive glucose analogue that gives a measure of glucose uptake and hence cellular metabolism, can be used to provide information about both tumour biology and tumour characteristics. In order to use this clinically, the PET-driven score needs to be further developed; however, better prognostics are not enough to improve outcome in the absence of effective therapeutic strategies. For example, a major point of contention is in the management of lymph nodes and whether there is benefit in systematic lymphadenectomy in this setting. If these PET-based measurements can predict lymph node metastasis more effectively than standard imaging techniques, they could guide which patients require lymphadenectomy and/or subsequent treatment.

A continuing multicentre trial, the ‘MAPPING study’, is currently evaluating the diagnostic accuracy of MRI and PET scans using both 18F-fluorodeoxyglucose and 18F-fluoro-l-thymidine (18F-FDG and 18F-FEC, respectively) at detecting lymph node metastasis in endometrial cancer (CRUK funded study, This will address the question of whether PET-based imaging of lymph nodes can reliably predict high-risk disease, and guide subsequent lymphadenectomy and/or adjuvant therapy. It is clear that imaging techniques that can assess tumour biology will provide valuable new insights to our pre-treatment investigations, enabling outcome prediction as well as treatment stratification.

Although this is a retrospective study with possible confounding factors, and the number of deaths in the cohort is too low to assess correlation between PET findings and overall survival, it does set the scene for further similar studies. Future studies using PET-based markers should specifically address their accuracy at identifying lymph node involvement and guiding the use of adjuvant therapy, ideally by incorporating them into interventional therapeutic trials in endometrial cancer. Only with clear evidence of efficacy can we justify using this imaging modality in routine clinical practice to predict tumour biology and modify treatment accordingly.

Disclosure of interests

The author has no conflicts of interest to declare.