Preoperative Factors Associated with Infiltrative Histologic Growth Patterns in Extremity Soft Tissue Sarcoma

Soft tissue sarcoma (STS) with an infiltrative histologic growth pattern, when compared to STS with an expansile pattern, may pose difficulties in local control. Preoperative assessment of the presence of infiltrative histologic growth pattern would be helpful in deciding treatment strategies. A review of 144 patients who underwent surgery for extremity STS was performed. Microscopically, the histologic growth pattern was defined as infiltrative if the penetration of the tumor cells into the surrounding tissue was observed. Possible clinicopathologic factors that might be associated with infiltrative histologic growth pattern were investigated with regard to patient demographics, tumor characteristics, and MRI findings. Of the 144 tumors, 71 (49%) showed infiltrative histologic growth pattern. On multivariate analysis, histological subtypes other than liposarcoma (OR = 4.57, p = 0.02) and infiltrative border on MRI (OR = 2.48, p = 0.01) were independent factors associated with infiltrative histologic growth pattern. Predictive index based on these two factors showed a significant improved accuracy (ROC-AUC = 0.647) for predicting infiltrative histologic growth pattern compared to either factor alone. Our data suggests that liposarcoma histology and tumor border on MRI can predict histologic growth pattern in extremity STS.


Introduction
Local recurrence in extremity soft tissue sarcomas (STS) not only is associated with poor oncologic outcome but also has deleterious effects on limb function [1][2][3][4]. Local recurrence is largely dictated by the ability to achieve histologically negative resection margins, and failure to obtain histologically negative margins represents microscopic residual disease and translates into high rates of local recurrence [5,6]. To obtain histologically negative resection margin, accurate determination of histologic tumor extent is necessary.
Histologic growth pattern has been associated with local recurrence in extremity STS [7][8][9]. STS with an infiltrative histologic growth pattern, when compared to STS with an expansile growth pattern, may pose difficulties in local control and subsequent risk for local recurrence [10,11]. Thus, preoperative assessment of the infiltrative histologic growth pattern, using preoperatively available clinicopathologic parameters, would be helpful in guiding surgery and adjuvant treatment. The purpose of this study was to identify clinicopathologic characteristics that are predictive of infiltrative histologic growth pattern in extremity STS. 144 patients for analysis. The institutional review board of our institute approved this study.

Assessment of Histologic Growth
Pattern. The cut surface of the whole tumor was obtained from the maximum diameter and was embedded into paraffin after which 4-m sections were made. The microscopic assessment was performed on 4-m sections stained with hematoxylin-eosin. The cut surface was examined, and several representative sections were made encompassing the tumor and surrounding tissue. Microscopically the histologic growth pattern was defined as infiltrative if the penetration of the tumor cells into the surrounding tissue was observed ( Figure 1(d)). Focal penetration of the tumor cells was regarded as infiltrative growth pattern. Histologic growth pattern was defined as expansile if the penetration of the tumor cells was not observed regardless of the presence of the pseudocapsule (Figure 1(a)) [10,12].

Statistical Analyses.
Continuous measures were compared using the independent-samples -test and categorical variables were compared using Pearson's chi-squared test. Clinical or radiological factors which were found to have a statistically significant association with the histologic growth pattern ( < 0.05) were included in a multivariate logistic regression analysis, with backward selection using the likelihood ratio test, to evaluate associations linking the histologic growth pattern. Statistical analysis was performed using SPSS v.21.0 software (IBM Inc., Armonk, New York). To test the predictive accuracy of identified variables, receiveroperating characteristic (ROC) curves were constructed, and the area under the curve (AUC) with a 95% confidence interval was calculated. Pairwise AUC comparisons were also performed between the two variables at using the nonparametric approach developed by DeLong et al. [22]. For all statistical comparisons, a value of 0.05 was considered statistically significant.  infiltrative histologic growth pattern were more likely to be superficial ( = 0.03) and to show infiltrative tumor border on MRI ( = 0.01) ( Table 2). Among the preoperative factors analyzed, there were no significant difference found with respect to age, gender, administration of preoperative radiation therapy or chemotherapy, tumor size, compartmental status, presentation status, or presence of peritumoral edema on MRI. STS with infiltrative histologic growth pattern showed trend towards having higher chance of pathologically positive margins and local recurrence ( Table 2).

Predictive Accuracy of Identified Factors.
To test the accuracy of predicting the presence of infiltrative histologic growth pattern, we generated a predictive index based on the presence or absence of two significant factors found in the multivariate analysis. Cases with both histological subtype other than liposarcoma and infiltrative tumor border on MRI were considered to have positive predictive index. All other cases were considered to have negative predictive index.

Discussion
Infiltrative histologic growth pattern has been associated with poor oncologic outcome, not only for local control but also for survival in extremity STS [7,8]. Preoperative assessment of infiltrative histologic growth pattern would be helpful in planning treatment of extremity STS. This study examined various preoperative clinicopathologic factors and identified nonliposarcoma histology and infiltrative tumor border on MRI as independent predictors of infiltrative histologic growth pattern. To our knowledge, this is the first study to examine various preoperative factors that can predict histologic growth pattern in extremity STS.
A few things should be considered while interpreting the results of this study. First, this study was based on retrospective review of a patient cohort from a single tertiary referral hospital with relatively small number of patients. The results of this study need to be validated in external databases in a prospective setting. Second, as for the evaluation of MRI characteristics, the MRI protocols were not uniform because of the retrospective nature of the study. Third, in our routine practice, the pathologic findings according to the specific radiological abnormalities were not evaluated, and such radiological-pathological correlation could not be performed due to the retrospective nature of the study. Fourth, some of the factors might not be accurately assessed preoperatively, such as the histological subtype or histological grade, especially in cases where core needle biopsy is used. Fifth, the limited use of recently developed biomarkers, such as MDM2, may hamper the accurate diagnosis of histological subtype.
In this study, larger tumor size was not associated with infiltrative histological growth pattern. The authors excluded well-differentiated liposarcomas, which often present with large tumor size, despite having a less aggressive biology. However, the association between tumor size and histological 6 Sarcoma growth pattern remained insignificant even after the exclusion of well-differentiated liposarcomas. Histological subtypes of STS seems to be associated with histologic growth pattern. Majority of STS subtypes showed predominantly infiltrative histologic growth pattern; UPS (62%, 16/26), myxofibrosarcoma (59%, 13/22), and synovial sarcoma (57%, 13/23). However, predominantly expansile histologic growth pattern was observed in MPNST (87%, 7/8) and liposarcoma (82%, 14/17). As current treatment strategies for STS are increasingly adapted to a specific histological subtype, predictive factors for specific subtypes of STS will be needed.
The value of tumor border on MRI in predicting histologic growth pattern seem to differ among histological subtypes of STS. The association of infiltrative tumor border on MRI, most notably the tail sign, with infiltrative histologic growth pattern has been well-documented in UPS and myxofibrosarcoma [7,[17][18][19][20]23]. Indeed, tumor border on MRI predicted infiltrative histologic growth pattern in UPS (AUC = 0.619, 95% CI = 0.385-0.852) and myxofibrosarcoma (AUC = 0.649, 95% CI = 0.391-0.907) with moderate accuracy in this study. However, for synovial sarcoma, ROC curve yielded an AUC of 0.519 (95% CI = 0.276-0.762). These data suggest that the predictive value of tumor border on MRI in assessing histologic growth pattern differs among histologic subtypes of STS.
Histologic growth patterns may be affected by the tumor environment. Tumors located in the superficial location, where no fascial boundaries exist, had more tumors with infiltrative histologic growth pattern than tumors located Sarcoma 7 in the deep location in the univariate analysis. Moreover, tumors presenting with previous unplanned surgery or local recurrence, in which the normal anatomical boundaries are violated, had more tumors with infiltrative histologic growth pattern than tumors presenting without any previous surgeries. Taken together, histologic growth pattern seems to be affected not only by the tumor biology but also by the tumor environment.
The predictive accuracy of the predictive index generated in this study was at best moderate with an AUC value of 0.647. Better predictive factors, such as genomic markers, are needed to improve predicting the histologic growth pattern in extremity STS. However, the results of this study can be implemented easily in the clinical practice and may provide a backbone for newly identified predictive factors of histologic growth pattern in extremity STS.

Conclusions
In conclusion, our data suggests that liposarcoma histology and tumor border on MRI can predict histologic growth pattern in extremity STS. If an extremity STS of nonliposarcoma histology shows infiltrative tumor border on MRI, infiltrative histologic growth pattern can be expected.

Additional Points
Our data suggest that patients with histological diagnosis other than liposarcoma and the presence of infiltrative border on MRI can predict infiltrative histologic growth pattern in extremity soft tissue sarcoma.

Disclosure
This paper was presented as a poster at The 19th International Society of Limb Salvage General Meeting.