Study population
56 patients who underwent surgery with the diagnosis of soft tissue sarcomas between June 2012 and March 2015 that were followed up with MRIs were evaluated.
After the surgery,
5 patients with positive surgical margins and residual disease,
5 patients with lung metastases was detected in and 2 patients with a history of a second primary cancer (one was colon cancer,
the other was papillary thyroid cancer) were excluded from the trial.
In addition,
two patients with a groin localized tumor and a hip prosthesis were excluded from the trial to avoid artefacts.
42 patients who underwent extensive resection and had clean surgical margins that attended their MRI follow ups at month 6 and later were included in the trial.
The study was approved by our institutional review board and all patients gave their informed consent.
The patient MRIs were reevaluated and compared to the results of the histopathological evaluation of the surgical specimens.
The demographic data and clinical findings of the patients were accessed using the patient archive system.
MRI Acquisition
All MRI examinations were performed on a 1.5T MR System (Signa Excite HD,
GE
Healthcare) with different coils knee,
shoulder,
body.
Axial and coronal T1weighted (spinecho imaging,
repetition time msec/ echo time msec = 580 700/9,428,3) and fluidsensitive (short tau inversion recovery imaging: 15005000/ 15142; T2weighted imaging: 35005700/15170) sequences,
as well as T1 weighted sequences (480/ 7,622,9,
8/1,5mmthick sections) after intravenous administration of contrast material (gadopentetate dimeglumine [Magnevist; Bayer Schering,
Berlin,
Germany],
0.1 mmol per kilogram of body weight),
were performed in all patients.
DWI was performed using an axial singleshot echoplanar imaging sequence centered on the lesions (b = 50,
400,
800,
and 1,000 s/mm2; 1,800/93.8; echotrain length,
2; bandwidth,
31,25 MHz; matrix size,
160 × 192; field of view,
2035 mm; number of signals averaged,
4; number of slices,
10; slice thickness,
7 mm; intersection gap,
1 mm; acquisition time,
2,5-3 minutes each for the four b values.
MRI Analysis and data collection
All images were transferred to a workstation (Advantage Windows version 4.2_07,
GE Healthcare) and the DWI sequence was postprocessed with commercial software (FuncTool,
GE Healthcare) to obtain ADC maps (blackandwhite and color,
the latter with a Puhthallium color scheme,
ranging from black,
diffusion restriction,
to red,
no diffusion restriction).
The ADC maps of each lesion were calculated using four b values (50,
400,800 and 1,000 s/mm2).
The scanner software provides the mean value within the ROI,
which equals the ADC value(multipliedby 10-3)
The MRI findings were reevaluated by radiologists with 8 years of experience (BS) and 10 years of experience (EA) in musculoskeletal MRI.
The signal characteristics,
contrast enhancement patterns,
locoregional distribution of the postoperative tissue areas and suspicious lesions were evaluated.
The hematomas,
seromas and areas of postsurgical or postradiotherapy soft tissue changes and the masses that may be relapses were identified based on the study by Garner and colleagues according to the T1,
T2 and contrasting patterns observed in the patients (6).
Lesions hyperintense in T1 weighted and T2 weighted images that did not demonstrate significant contrast enhancement following IV contrast administration were classified as hematomas; lesions with hypointense signal characteristics in T1 weighted images and hyperintense in T2 weighted images that did not enhance with contrast after IV contrast administration or contrasted peripherally as seromas; areas with hypointense signal characteristics in T1 weighted images and hyperintense in T2 weighted images that were mildly enhanced with contrast after IV contrast administration as changes after operation and radiotherapy ; and lesions that were T1 hypointense,
T2 hyperintense and enhanced with contrast after IV contrast administration (Figures1,2,3) were classified as potential recurrent masses.
In diffusion weighted images,
1cm2 ROIs were placed in postoperative tissue areas that were hyperintense in T2 weighted images and in 3 different areas with an appearance suspicious of recurrent masses and the mean ADC values were calculated.
ADC calculation was not done in the areas that were hypointense in T1T2 weighted images and considered as fibrosis.
After MRI examinations,
biopsies were obtained from the suspicious lesions with wire localization biopsies in 2 patients and 13 patients underwent total excision.
The mean diameter of the lesions totally excised was measured as 4,3±1,5 cm (range 28 cm).
Patients regarded as posttreatment tissue changes according to the T1,
T2 and contrasting characteristics were followed up on for a mean period of 17,8±6,7 months (range,
min 8 – max 32 months).
No relapse or change in the tissue areas was identified during follow up.
Pathology:
1 patient underwent surgery for an alveolar rhabdomyosarcoma,
1 for an embryonal rhabdomyosarcoma,
8 for fibrosarcomas,
6 for leiomyosarcomas,
8 for pleomorphic liposarcomas,
6 for myxoid liposarcomas,
2 for angiosarcomas,
1 for a malign schwannoma,
4 for pleomorphic sarcomas,
1 for a myxofibrosarcoma,
3 for synovial sarcomas and 1 for a soft tissue osteosarcoma.
Among the tumors that relapsed,
1 was an alveolar rhabdomyosarcoma,
3 were fibrosarcomas,
2 were leiomyosarcomas,
3 were pleomorphic liposarcomas,
1 was a myxoid liposarcoma,
2 were angiosarcomas,
1 was a malignant schwannoma,
1 was a synovial sarcoma and 1 was a soft tissue osteosarcoma.
5(11,9%) of sarcomas were grade1,
21(50%) were gared 2 and 16 (38,1%) were grade 3 tumors.
Statistical analysis
Descriptive statistics were expressed as mean±satandart deviation for continuous variables count and percentage for categorical variables.
Chi-square test was used to compare difference between groups for categorical variables.
Repeated measures ANOVA was performed to compare means of different b values (b 50,
b400,
b800,
b1000).
A post hoc Bonferroni test was used to examine for multiple comparisons.
The discriminatory performance of a quantative test (b 50,
b 40,
b 800,
b 1000) was evaluated by receiver-operating characteristics(ROC) analysis.
The ROC curve was plotted based on the true positive rate (TPR) and false positive arte(FPR).
The area under the ROC curve (AUC) was obtained using an on parametric method (trapezoidal rule).
Histopathologically and the MR follow up was taken as the reference standart method.
Areas under the ROC curves were compared for b 50,
b 400,
b 800,
b 1000.
Results of statistical significance were provided after Bonferroni’s correction.
After ROC analysis,
thr optimal cutt-off point was determined according to the Youden index.
Sensivity,
specifity,
positive predictive value(PPV) and negative predictive value(NPV)for b values were calculated according to optimal cutt-off point A p< 0.05 was accepted as statistically signifcant.
All statistically analyses were performed by using IBM SPS Statistics 21.0 software.