In our study we investigated the diagnostic accuracy of standard MRI for detecting recurring soft-tissue sarcoma at a multidisciplinary sarcoma center. Soft-tissue tumors include benign and malignant tumors and tumor-like lesions . In our study we dealt with malignant lesions only, so-called soft-tissue sarcomas (STS). STS constitute a rare and heterogeneous group of tumors, which account for only about 1% of all malignancies . The most recent histological classification of STS comes from the current WHO histological typing, in which more than 50 subtypes are described [1, 15, 16]. Previous studies on soft-tissue sarcoma recurrence have reported a wide range of recurrence rates of up to 50%[17, 18]. Due to the rarity of STS, imaging studies on STS still remain scarce. Furthermore, most previous studies dealt with imaging features rather than with the diagnostic accuracy. Additionally, previous diagnostic studies mostly included small patient numbers or reviewed the available literature. For postoperative surveillance, MRI is the imaging modality of choice and is widely used to assess STS recurrence [14, 19], as MRI has the advantage of a high soft-tissue contrast and no radiation. Nevertheless, differentiation between post-treatment changes and recurrent STS on MRI is often reported to be challenging . Previous publications have often shown that post-treatment changes and scar tissue can obscure recurring STS, which leads to unnecessary biopsies . In patients with local recurrences combined treatment with surgery and additional radiation therapy is usually chosen to improve local control [1, 22]. Nevertheless, a decision regarding the use of additional radiation therapy should be evaluated from case to case [1, 23]. Before starting the therapy, core needle biopsy is often performed to identify the pathology of the suspected lesions . To start therapy quickly and to avoid unnecessary biopsy, precise post-operative MRI diagnostics are indispensable. Therefore, according to our study, it is of high clinical and diagnostic importance to determine the diagnostic reliability of MRI for postoperative surveillance of sarcoma patients in a real-world cohort over a long time period. In our study, both sensitivity (92%) and specificity (98%) were high overall, even after both surgery and radiation therapy and in both subcutaneous and intramuscular lesions. Indeed, the sensitivity was lower after radiation therapy than after surgery alone and in subcutaneous lesions, compared to intramuscular lesions, but the sensitivity still remained at a high level. Reasons for the decreased sensitivities could lie in the increased rate of soft-tissue alterations after additional radiation therapy[21, 25] and the usually smaller sizes of the subcutaneous lesions. These two findings ultimately lead to more difficulty in distinguishing between post-treatment changes and recurring tumor. The range of the described sensitivity and specificity is high. However, some previous publications reported a lower specificity, ultimately leading to unneeded biopsies [14, 21, 26]. Afonso et al. described a sensitivity and specificity of only 58% and 73%, respectively, for conventional MRI , while Del Grande F. et al. assessed a sensitivity and specificity of MRI of 100% and 52%, respectively, in detecting tumor recurrence in nonenhanced MRI, and 100% and 97%, respectively, in contrast-enhanced MRI . In a recent review, Pennington A. et al. calculated a mean sensitivity and specificity of 88% and 86%, respectively, for local recurrences of primary vertebral tumors. Other authors showed sensitivities and specificities of 64–88% and 85–96%, respectively, for MRI for detecting soft-tissue tumors [19, 28–31]. Nevertheless, none of the previous studies investigated how MRI performed in the real-world clinical setting. Previous publications reported a lack of specificity of MRI for detecting recurring STS in nonenhanced T1- and T2-weighted images [21, 32, 33]. Therefore, contrast-enhanced MRI is reported to improve the diagnostic accuracy of MRI .
In our study, all of the patients were examined using contrast-enhanced MRI. Our data emphasize that MRI is a highly valuable imaging modality in the long-term postoperative surveillance of STS patients. Nevertheless, we found 23 false-positive and 6 false-negative results (8.5%). These 6 cases were all derived from streaky and small ovoid/nodular lesions, which were difficult to distinguish from the surrounding post-treatment tissue. A recent study showed that distinct post-therapeutic changes are the main reason for false-negative results on MRI . All false-positive results were found in patients in whom the primary STS was polycyclic/multilobulated. This may well be due to the fact that polycyclic/multilobulated primary tumors are larger than other configured STS in the mean and therefore perhaps lead to more heterogeneous post-treatment tissue changes. Furthermore, 22 of the 23 false-positive results were derived from patients with R0 resection. This is in contrast to a recent study showing that microscopic positive margins were the main reason for false-positive results .
Therefore, radiologists should pay particular attention to patients in whom the primary STS was polycyclic/multilobulated in shape and should carefully screen the soft tissue for streaky or small ovoid/nodular recurrences. However, this fact also demonstrates the limitations of MRI surveillance. In cases of recurrent lesions with a a grainy appearance, it may become difficult to correctly detect a recurring tumor in the surrounding tissue. Therefore, in suspected or unclear cases, the subsequent MRI follow-up examinations should take place after 3 to 6 months.
Our study has some limitations, as it is a single-center study with a retrospective design. Nevertheless, we could include 204 patients in a 12-year survey with a total of 1055 MRI follow-up scans. Another limitation is the verification of the true-positive/-negative and false-positive/-negative results. Biopsy was only performed in patients in whom recurrence was suspected. In the other cases, radiologists evaluated whether recurrences were overlooked or not. Therefore, we cannot completely rule out that individual findings might be false.