Osteosarcoma is a malignant mesenchymal neoplasm

Transcription

1 Original Articles Histopathologic Features of Prognostic Significance in High-Grade Osteosarcoma Michael Herman Chui, MD; Rita A. Kandel, MD; Marcus Wong, BSc; Anthony M. Griffin, MSc; Robert S. Bell, MD; Martin E. Blackstein, MD, PhD; Jay S. Wunder, MD, MSc; Brendan C. Dickson, MD, MSc Context. In osteosarcoma treated with neoadjuvant chemotherapy the extent of tumor necrosis on resection is considered an indicator of treatment response, and this has been shown to correlate with survival in most but not all studies. Objective. To identify additional histologic variables of prognostic significance in high-grade osteosarcoma. Design. Slides of pretreatment biopsy and primary postneoadjuvant chemotherapy resections from 165 patients with high-grade osteosarcoma were reviewed. Univariate (Kaplan-Meier) and multivariate (Cox regression) analyses were performed to identify clinical and histomorphologic attributes associated with overall survival. Results. Univariate analyses confirmed the prognostic significance of metastatic status on presentation, primary tumor size, anatomic site, and histologic subtype. Additionally, the identification of lymphovascular invasion, 10% or more residual viable tumor, and 10 or more mitoses per 10 high-powered fields assessed in posttreatment resections were associated with poor survival, retaining significance in multivariate analyses. Based on results from multivariate analysis, we developed a prognostic index incorporating primary tumor size and site, and significant histologic features assessed on resection (ie, lymphovascular invasion status, mitotic rate, and extent of viable tumor). This scoring system segregates patients into 3 risk categories with significant differences in overall survival and retained significance in an independent validation set of 42 cases. Conclusions. The integration of clinical and microscopic features improves prognostication of patients with osteosarcoma. (Arch Pathol Lab Med. 2016;140: ; doi: /arpa OA) Accepted for publication February 29, Published as an Early Online Release August 23, Supplemental digital content is available for this article at www. archivesofpathology.org in the November 2016 table of contents. From the Departments of Pathology and Laboratory Medicine (Drs Chui, Kandel, and Dickson, and Mr Wong), Medicine (Dr Blackstein), and Orthopedic Surgery (Mr Griffin and Dr Wunder), Mount Sinai Hospital, Toronto, Ontario, Canada; and the Departments of Laboratory Medicine and Pathobiology (Drs Chui, Kandel, and Dickson), Orthopedic Surgery (Drs Bell and Wunder), and Medical Oncology (Dr Blackstein), Faculty of Medicine, University of Toronto, Toronto, Ontario, Canada. The authors have no relevant financial interest in the products or companies described in this article. Reprints: Brendan C. Dickson, MD, Department of Pathology and Laboratory Medicine, Mount Sinai Hospital, 600 University Ave, Rm , Toronto, ON M5G 1X5, Canada ( bdickson@ mtsinai.on.ca). Osteosarcoma is a malignant mesenchymal neoplasm characterized by bone production. 1 Despite representing the most common primary bone sarcoma, the overall incidence is comparatively low. 2 Several types of primary high-grade osteosarcoma are recognized by the World Health Organization, including the conventional types encompassed by the chondroblastic, fibroblastic, and osteoblastic subtypes, among others. 3 Despite heterogeneity in histomorphology, most high-grade osteosarcomas receive similar treatment, namely, neoadjuvant chemotherapy typically consisting of doxorubicin, cisplatin, and highdose methotrexate, followed by surgical resection and maintenance chemotherapy. 4 Histopathologic prognostic factors represent an essential component to the treatment stratification of many types of cancer. Although a number of groups have assessed the contribution of clinical parameters namely, patient age, sex, anatomic site of primary tumor, tumor size, and metastasis at presentation in predicting survival outcomes for patients with osteosarcoma, 5 9 there are limited data concerning histopathologic attributes. Theroleoftumorpleomorphism, 10 mitotic activity, 11 and morphologic subtype 6,10,12 14 as prognostic factors appears to be limited. Moreover, it is unclear whether these features should be assessed on treatment-naive biopsy specimens or on resection following neoadjuvant chemotherapy. Although the extent of chemotherapy-induced tumor necrosis has been shown to be a strong prognostic factor, 7,8,15,16 several studies have demonstrated contradictory results, 4,9,17 emphasizing a need for standardization in the method of histologic assessment of treatment response. 18 In this study, we reviewed matched pretreatment biopsy and postneoadjuvant chemotherapy surgical resection specimensofhigh-gradeosteosarcomatoidentifyhistomorphologic features of prognostic significance. We also sought to determine whether the combination of clinical and pathologic features could enable more robust stratification of patients with significantly different survival outcomes. Arch Pathol Lab Med Vol 140, November 2016 Prognostic Factors in Osteosarcoma Chui et al 1231

2 Feature Table 1. Patient Characteristics Derivation Cohort, No. (%) of Patients (n ¼ 165) Validation Cohort, No. (%) of Patients (n ¼ 42) P Value Age.07,40 y 129 (78) 27 (64) 40 y 36 (22) 15 (36) Sex.72 Male 103 (62) 28 (67) Female 62 (38) 14 (33) Size cm 67 (41) 17 (40).8.0 cm 98 (59) 25 (60) Metastasis at presentation.83 No 129 (78) 34 (81) Yes 36 (22) 8 (19) Clinical stage (AJCC 7th edition) IIA 57 (34) 15 (36) IIB 72 (44) 19 (45) III/IV 36 (22) 8 (19) Location Trunk 23 (14) 7 (17).23 Distal extremity 115 (70) 24 (57) Proximal extremity 27 (16) 11 (26) Secondary to preexisting bone condition a.24 No 158 (96) 38 (90) Yes 7 (4) 4 (10) Margin status..99 Negative 155 (94) 40 (95) Positive 10 (6) 2 (5) Abbreviation: AJCC, American Joint Committee on Cancer. a In the derivation cohort, these included Paget disease (1 case), bone infarct (2 cases), and radiation induced (4 cases). In the validation set, these included fibrous dysplasia (1 case), giant cell tumor (1 case), and radiation induced (2 cases). MATERIALS AND METHODS Patient Selection Following institutional Research Ethics Board approval, a retrospective pathology review was performed of the biopsies and/or primary resection specimens from patients treated at our institution for high-grade osteosarcoma. Clinical data were obtained from patient charts and electronic medical records. Between 1988 and 2011, a total of 207 patients underwent surgery for high-grade osteosarcoma at our institution. Slides from both biopsy and resection specimens were available in 165 cases, which constituted our derivation cohort. Slides of resection specimens alone were available in an additional 42 cases, which comprised the validation cohort for subsequent analyses. Histology Review Sampling of the resection specimens included, at minimum, one full cut surface of the tumor and sampling of all relevant margins. In each case, all available slides were examined by reviewers blinded to clinical outcome (initial review and scoring by M.H.C.; confirmatory review by B.C.D.). Tumor subtype was assigned according to terminology and criteria defined in the World Health Organization Classification for Bone and Soft Tissue Tumors, 3 with the exception of chondroblastic subtype, which we defined as any high-grade osteosarcoma with at least 5% of the tumor containing neoplastic chondroid matrix. We applied a lower threshold for this designation because we hypothesized that the presence of any malignant chondroid might impact prognosis, in light of previous work demonstrating chondroblastic subtype to be associated with poor response to chemotherapy. 19 Lymphovascular invasion (LVI) was considered present when tumor cells were identified within an endotheliallined space. Mitotic counts were obtained from 10 successive highpower fields (HPFs) in the most mitotically active area of tumor (3400; field diameter, 0.55 mm). Nuclear pleomorphism was evaluated in the area with the most severe nuclear atypia at intermediate magnification (3100), according to criteria adapted from the Nottingham grading system applied in breast carcinoma 20 ; namely, uniform nuclei with minimal variability in size and shape, moderate variation in size and shape, and marked nuclear variation. Other histologic parameters included tumor viability, tumor necrosis, osteoid matrix, and fibrosis/hyalinization. These features were quantified by estimating the percentage of each parameter in each tumor slide and calculating the mean based on the total number of tumor slides examined. Tumor necrosis was defined as areas occupied by ghost cells, cellular debris, and empty lacunar spaces, and excluded areas of hemorrhage, fibrosis, and extracellular matrix. Statistical Analysis Correlations between various clinical and pathologic features were assessed using the Mann-Whitney U test for continuous variables and the Fisher exact test for categoric variables, as applicable. The j concordance test was used to assess the concordance of tumor subtyping between biopsy and resection specimens. Overall survival was determined from the date of primary surgery to the date of death from any cause or date of last followup. The relationships between survival and possible prognostic factors were evaluated by Kaplan-Meier plots and log-rank tests. Clinical and histologic features significant in univariate analyses were entered in a Cox proportional hazards model, reduced by backward elimination (likelihood ratio) to retain only predictors significant at P,.10. All continuous variables were categorized; cutpoints were defined a priori, according to values defined in previous studies (age, 7,21 tumor size, 22 mitotic rate, 23 necrosis, residual viable tumor after treatment 7,15,26 ) and/or at conceptually meaningful values (osteoid, fibrosis). The proportionality assumption was checked by visual inspection of cumulative hazard and Martingale residual plots. Given the exploratory nature of the 1232 Arch Pathol Lab Med Vol 140, November 2016 Prognostic Factors in Osteosarcoma Chui et al

3 Figure 1. Histologic alterations associated with chemotherapy effect in high-grade osteosarcoma. A, Degenerating ghostlike cells set within osteoid/chondroid matrix. B, Sheets of sclerotic osteoid matrix. C and D, Necrosis. E, Hemorrhagic degeneration. F, Fibrosis/hyalinization (hematoxylin-eosin, original magnifications 3100 [A, B, and F], 3200 [C and D], and 340 [E]). Arch Pathol Lab Med Vol 140, November 2016 Prognostic Factors in Osteosarcoma Chui et al 1233

4 Table 2. Feature Univariate Analysis of Clinical and Histopathologic Features With Overall Survival No. (%) of Patients 5-Year Overall 10-Year Overall P Value (Log-Rank) Clinical Age.10,40 y 129 (78) y 36 (22) Sex.47 Female 103 (62) Male 62 (38) Size 8.0 cm 67 (41) cm 98 (59) Metastasis at presentation,.001 No 129 (78) Yes 36 (22) Clinical stage (AJCC 7th edition) 22,.001 IIA 57 (34) IIB 72 (44) III/IV 36 (22) Location,.001 Trunk 23 (14) Distal extremity 115 (70) Proximal extremity 27 (16) Histopathology Tumor subtype Osteoblastic 41 (25) Fibroblastic 44 (27) Chondroblastic 60 (36) Telangiectatic 17 (10) Other a 3 (2) Nuclear pleomorphism Biopsy I 4 (2) II 70 (43) III 91 (55) Resection I 2 (1) II 39 (28) III 101 (71) Mitotic rate (per 10 HPF) Biopsy (50) (22) (28) Resection (66) , (20) (14) Lymphovascular invasion Biopsy Present 42 (26) Not identified 123 (74) Resection Present 44 (27) ,.001 Not identified 121 (73) Necrosis Biopsy Present 123 (75) Not identified 42 (25) Resection,50% 128 (78) % 37 (22) Osteoid Biopsy 0% 9% 78 (47) % 49% 73 (44) % 14 (8) Resection 0% 9% 83 (50) % 49% 67 (41) % 15 (9) , Arch Pathol Lab Med Vol 140, November 2016 Prognostic Factors in Osteosarcoma Chui et al

5 Feature Table 2. No. (%) of Patients Continued 5-Year Overall 10-Year Overall P Value (Log-Rank) Fibrosis/hyalinization Biopsy 0% 9% 126 (76) % 25% 39 (24) Resection 0% 9% 54 (33) % 49% 85 (51) % 26 (16) Tumor viability after chemotherapy (resection) 0% 9% 66 (40) , % 49% 62 (38) % 37 (22) Margins (resection) Negative 155 (94) Positive 10 (6) Abbreviations: AJCC, American Joint Committee on Cancer; HPF, high-power field (3400; field diameter, 0.55 mm). a Includes small cell (n ¼ 1), epithelioid (n ¼ 1), and giant cell rich (n ¼ 1). analyses, P values were not adjusted for multiple comparisons. Statistical significance was defined at P,.05. The predictive power of prognostic models was evaluated using Harrell C statistic (for which a value approaching 1 is indicative of better discrimination). RESULTS Patient Demographics and Clinical Characteristics Within the derivation cohort, the median age at diagnosis was 22 years (range, years). Median tumor size at resection was 9.0 cm (range, cm). In 7 cases, osteosarcoma was secondary to a preexisting bone condition: Paget disease (n ¼ 1; 14%), bone infarct (n ¼ 2; 29%), and radiation-induced (n ¼ 4; 57%); these cases occurred in an older age group (median age, 43 years; range, years). Primary tumor sites were categorized as previously described 7 : trunk (acetabulum [n ¼ 2], ilium [n ¼ 17], pubic ramus [n ¼ 3], and clavicle [n ¼ 1]), proximal extremity (upper two-thirds of femur [n ¼ 13] and upper two-thirds of humerus [n ¼ 14]), and distal extremity (tibia [n ¼ 29], fibula [n ¼ 10], distal femur [n ¼ 65], distal humerus [n ¼ 3], carpus [n ¼ 1], and distal radius [n ¼ 7]). The more centrally located tumors (ie, trunk or proximal limb) were significantly associated with older age (median, 32 versus 21 years; P ¼.003), larger tumor size (median, 12.0 versus 8.5 cm; P ¼.007), and metastases at presentation (17 of 50 [34%] versus 19 of 115 [17%]; P ¼.02). Median tumor sizes for localized and metastatic disease at presentation were 8.5 cm and 11.0 cm, respectively (P ¼.003). No significant associations were found between primary metastases and age, sex, or histologic subtype. The presence of tumor at the surgical resection margin was identified in 10 of 165 patients (6.1%; 3 grossly positive and 7 microscopic). All patients were treated with standard neoadjuvant chemotherapy. Comparison of patient characteristics between derivation and validation cohorts did not reveal any significant differences (Table 1). Histologic Features Tumor subtype classification was assigned on both biopsy and resection specimens independently. Overall, there was a high level of concordance between the initial and final diagnoses (j ¼.89; Supplemental Table 1; see supplemental digital content, containing 3 tables at www. archivesofpathology.org in the November 2016 table of contents). There were no significant associations between tumor subtype and age, sex, tumor site, or stage. Various histologic alterations associated with treatment effect were observed and were consistent with previously published descriptions (Figure 1) An admixture of degenerating wispy osteoid matrix with empty lacunae, necrotic ghost cells, and edematous stroma with loose fibrous tissue was a commonly observed pattern of response. Other areas showed dense fibrosis or hyalinization, particularly prominent in fibroblastic osteosarcomas, whereas replacement of tumor by sheets of sclerotic acellular osteoid matrix was most common in osteoblastic osteosarcomas. Survival Analysis For the cohort of 165 patients, median overall survival was 49 months (interquartile range, months; range, months), and 5-year overall survival was 52% 6 4%. At last follow-up, 79 patients were alive with no evidence of disease, 7 were alive with disease, 74 died of the disease, and 5 died of other causes (cirrhosis [n ¼ 1], pulmonary embolism [n ¼ 1], massive hemorrhage from surgery [n ¼ 1], and multiple organ dysfunction [n ¼ 2]). By univariate analysis (Table 2), clinical parameters associated with all-cause mortality included tumor size, metastatic status, clinical stage, and tumor site. With respect to histologic parameters assessable on initial biopsy, only tumor subtype achieved statistical significance, with chondroblastic subtype associated with shorter overall survival (Figure 2, A). Spontaneous tumor necrosis at initial biopsy was of borderline significance (Figure 2, B). In resection specimens, percent of residual viable tumor, mitotic activity, LVI, and percent of fibrosis/hyalinization were significantly associated with overall survival (Figures 2, C through F, and 3). Of note, the year of diagnosis did not correlate with overall survival when analyzed either as a continuous variable (hazard ratio, 1.01 [CI, ]; P ¼.60) or when comparing cases receiving a diagnosis prior to the year 2000 (n ¼ 85) with those receiving a diagnosis after (n ¼ 80; P ¼.55), thus supporting our initial decision to not exclude older cases. Arch Pathol Lab Med Vol 140, November 2016 Prognostic Factors in Osteosarcoma Chui et al 1235

6 Figure 2. Histologic features of prognostic significance (univariate analysis). Kaplan-Meier curves demonstrating overall survival with (A) tumor subtype, as determined from both biopsy and resection specimens; (B) necrosis, detected in biopsy specimen; (C) lymphovascular invasion; (D) mitotic rate; (E) residual tumor viability; and (F) extent of fibrosis. C through F are as assessed following surgical resection. Abbreviation: HPF, highpowered field (field diameter ¼ 0.55 mm) Arch Pathol Lab Med Vol 140, November 2016 Prognostic Factors in Osteosarcoma Chui et al

7 Table 3. Multivariate Analysis of Clinical and Histopathologic Features Associated With Overall Survival a Characteristic Hazard Ratio 95% CI P Value Age.06,40 y 1 40 y Size, cm cm Metastasis.05 No 1 Yes Location.02 Distal extremity 1 Proximal extremity Trunk Lymphovascular invasion.01 Not identified 1 Present Tumor viability.02 0% 9% 1 10% 49% % Mitotic rate, a Backward stepwise selection, retaining significant factors at P,.10, was performed on the following covariates entered into the Cox model: age, tumor size, location, and subtype, presence of necrosis on biopsy, and metastatic status; and, as assessed at resection, lymphovascular invasion, extent of fibrosis, residual tumor viability, mitotic rate, chondroid, and margin status. By multivariate analysis, considering only factors assessable at the time of diagnosis, patient age, tumor size, site, and subtype retained significance as independent prognostic factors (Supplemental Table 2). In the multivariate model that also included treatment-related changes, only tumor size and site, LVI, posttreatment mitotic rate, and residual tumor viability retained independent prognostic significance (Table 3). Notably, for posttreatment mitotic rate and Table 4. Proposed Scoring Criteria for an Integrated Prognostic Index Parameter Score Tumor size 8.0 cm cm 1 Lymphovascular invasion Not identified 0 Positive 1 Tumor location a Distal extremity 0 Other site 1 Residual tumor viability,10% 0 10% 1 Mitotic rate,10 per 10 HPF 0 10 per 10 HPF 1 Prognostic index Low risk Sum ¼ 0or1 Intermediate risk Sum ¼ 2or3 High risk Sum ¼ 4or5 Abbreviation: HPF, high-power fields (3400; field diameter, 0.55 mm). a Distal extremity includes any location distal to the upper two-thirds of the humerus or femur. residual tumor viability, hazard ratios corresponding to the intermediate and highest factor levels were similar in magnitude, as were the 5-year overall survival rates. Similarly, in comparison with distal extremity tumors, survival outcomes for tumors located in the trunk or proximal extremity appeared to be just as poor. Performing stepwise multivariate analysis with these parameters collapsed into binary categoric variables retained the same predictors in the Cox model. Development of an Integrated Prognostic Index Given the independent prognostic significance of tumor size and site, LVI, residual tumor viability, and mitotic rate, we postulated that combining these features would provide a more accurate prediction of survival outcomes. Each Figure 3. Viable tumor cells after neoadjuvant chemotherapy. A, Degenerating tumor cells with no mitotic activity. At the time of last follow-up (62 months), the patient remained alive with no evidence of disease. B, Chondroblastic osteosarcoma with conspicuous mitotic figures; associated with overall survival of 8 months (hematoxylin-eosin, original magnifications 3200 [A] and 3400 [B]). Arch Pathol Lab Med Vol 140, November 2016 Prognostic Factors in Osteosarcoma Chui et al 1237

8 Figure 4. Kaplan-Meier curves for derivation and validation patient cohorts demonstrating overall survival separated by (A and B) residual tumor viability and by (C and D) integrated prognostic index. parameter was given equal weight because hazard ratios were comparable in magnitude. The scoring criteria are summarized in Table 4. Compared with residual tumor viability alone, the proposed prognostic index demonstrated improved stratification of prognostic groups, enabling separation of survival curves into low-, intermediate-, and high-risk categories (Figure 4; Table 5). The prognostic significance of this scoring system was confirmed in an independent set of 42 high-grade osteosarcomas. Moreover, Harrell C statistic, a quantitative measure of the predictive accuracy of a prognostic factor, is consistently higher for the integrated prognostic index than for tumor viability alone Table 5. Survival Outcomes According to Residual Tumor Viability and Integrated Prognostic Index No. (%) Derivation Cohort (n ¼ 165) 5-Year Overall 10-Year Overall P Value (Log-Rank) Hazard Ratio (95% CI) Residual tumor viability,10% 66 (40) , % 99 (60) ( ) C statistic.62 Integrated prognostic index Low (0 or 1) 61 (37) , Intermediate (2 or 3) 77 (47) ( ) High (4 or 5) 27 (16) ( ) C statistic.73 Pairwise comparisons Low versus intermediate, P,.001 Low versus high, P,.001 Int. versus high, P, Arch Pathol Lab Med Vol 140, November 2016 Prognostic Factors in Osteosarcoma Chui et al

9 (.73 versus.62 [derivation cohort],.75 versus.65 [validation cohort]). Considering the entire data set of 207 patients (derivation and validation sets combined), 61 cases (29%) were classified as good prognosis/low risk under the proposed scoring system and by assessment of tumor viability alone (ie,,10%), whereas 34 cases (16%) were classified as poor prognosis/high risk by both systems (Supplemental Table 3). Both intermediate- and high-risk groups defined by the integrated prognostic index are composed of a large proportion of cases with high residual tumor burden. However, mitotic counts 10 or greater per HPF and LVI at surgical resection were more frequently found in cases designated as high risk (Figure 5). DISCUSSION Osteosarcoma is rare, and there are few studies exploring the prognostic significance of microscopic tumor attributes. To our knowledge this is the first study to systematically assess a variety of histopathology-derived parameters both before and after neoadjuvant chemotherapy in a clinically annotated cohort of high-grade osteosarcomas. Because neoadjuvant chemotherapy represents a routine part of the standard treatment regimen, the only opportunity to evaluate unadulterated tumor histology is from the diagnostic tissue biopsy. These samples are inherently limited in size and therefore prone to sampling error because the material may not accurately reflect the overall tumor. Therefore, unsurprisingly, none of the features typically associated with aggressive behavior (ie, high mitotic rate, LVI, nuclear pleomorphism) were found to significantly correlate with survival in our pretreatment cohort, with exception of a weak relationship observed with necrosis. In a study conducted prior to the widespread use of neoadjuvant chemotherapy in the treatment of osteosarcoma, spontaneous tumor necrosis was found to be an independent prognostic factor, even after adjusting for tumor size. 30 The reported stronger association may be attributable to more accurate estimates of necrosis from the examination of surgical resections rather than biopsies. Consistent with prior reports, tumor size, 7,14,31,32 anatomic site, 7,13,15,32 and metastatic status at presentation 7,10,31,33,34 were strong determinants of survival outcome. After adjusting for these parameters, tumor subtype retained independent prognostic significance. The prognostic significance of chondroblastic differentiation has been controversial, with conflicting reports of association with better 12 as well as worse 5,33 survival, whereas fibroblastic tumors have been associated with better prognosis compared with osteoblastic and chondroblastic subtypes. 13 Such discrepancies may be partially explained by differences in diagnostic criteria employed in previous studies. According to Dahlin and Unni s classification, 34 subsequently expanded by the World Health Organization, the subclassification of conventional osteosarcomas is based on the predominant line of differentiation. Although other studies have applied thresholds of 30% 17 to 50% 13 chondroid matrix for diagnosing chondroblastic osteosarcoma, our choice of 5% appears to show prognostic significance by univariate analysis. Notably, in the multivariate model incorporating treatment-related parameters, subtype was not retained as an independent prognostic factor, suggesting that the association is dependent on treatment response. Further studies are needed to standardize criteria for histologic subtyping of osteosarcomas, because accumulating data suggest this may represent a promising predictive marker obtainable by means of targeted biopsy sampling, which may eventually influence treatment planning. Traditionally, response to chemotherapy was described in terms of tumor necrosis; similar to other investigators, we prefer a definition based on percent viable tumor cells. 26,28,35 Such a distinction encompasses, in addition to tumor necrosis, the presence of fibrosis and acellular matrix as elements of nonviable tumor, and cumulatively represents treatment effect. In routine practice, a pathologist would presumably consider the entire area occupied by necrotic cells, acellular matrix, and loose fibrous tissue in estimating the extent of necrosis. We chose to separate these individual components to assess their relative importance in representing response to chemotherapy. Interestingly, the only element that demonstrated a statistically significant correlation with overall survival, albeit only in univariate analysis, was fibrosis/hyalinization. Consistent with previous reports, 7,8,14 16,36,37 posttreatment tumor viability was an important determinant of overall survival. However, in several studies, good histologic response, as defined by the extent of necrosis, did not translate into survival benefit. 4,9,17 Our results support the routine estimation of posttreatment tumor viability, particularly emphasizing the importance of considering fibrosis when assessing treatment response. Although pretreatment mitotic activity failed to show a relationship with prognosis, 10,11 we report a novel associa- Table 5. Extended Validation Cohort (n ¼ 42) No. (%) 5-Year Overall 10-Year Overall P Value (Log-Rank) Hazard Ratio (95% CI) 14 (33) (67) ( ) (38) , (43) ( ) 8 (19) ( ).75 Low versus intermediate, P ¼.001 Low versus high, P,.001 Intermediate versus high, P ¼.09 Arch Pathol Lab Med Vol 140, November 2016 Prognostic Factors in Osteosarcoma Chui et al 1239

10 Figure 5. Frequency distributions segregated by prognostic index risk category for (A) residual tumor viability (,10% and 10%), (B) primary tumor site (axial site, proximal limb, distal limb), (C) tumor size (8 cm and.8 cm), (D) mitotic rate (,10 per HPF and 10 per HPF), and (E) lymphovascular invasion (LVI; not identified and present). Abbreviation: HPF, high-powered field (field diameter ¼ 0.55 mm) Arch Pathol Lab Med Vol 140, November 2016 Prognostic Factors in Osteosarcoma Chui et al

11 tion between posttreatment mitotic activity and overall survival in osteosarcoma. Previous studies in breast cancer reported an association between high Ki-67 index after neoadjuvant chemotherapy and poor prognosis. 38 The presence of quiescent viable tumor cells, often with degenerative changes, likely represents treatment-induced tumor dormancy. In contrast, mitotically active tumor cells may represent resistance to cytotoxic therapy, and thus are relevant to the evaluation of treatment effect. Multivariate nomograms, based on Cox regression, were recently developed for predicting distant metastases and overall survival in patients with osteosarcoma. 39 Similar to previous outcome-related studies, this investigation primarily focused on clinical attributes and reported an association with patient age, sex, tumor size, tumor site, and the presence of pathologic fracture. This study additionally confirmed, on multivariate analysis, that histopathologic assessment of the degree of tumor necrosis was significantly associated with the distant metastases and overall survival. An important factor that distinguishes our work from previous studies is the extent of histomorphologic attributes examined, as well as the fact that we considered findings in both the preadjuvant and postadjuvant settings. Because of disease rarity and the practical limitations in obtaining sufficient numbers of cases fulfilling the inclusion criteria, a limitation of this study is the fact that a power analysis was not performed to determine the optimal size of the validation cohort. With the available, albeit limited, sample of 42 cases, no statistical differences were detected between all pairwise comparisons. This analysis nevertheless serves as a proof of principle, and validation of the proposed scoring systems in larger data sets is necessary in future studies. In summary, our findings suggest that detailed microscopic examination yields useful prognostic information in high-grade osteosarcoma. In biopsy samples, necrosis and chondroblastic differentiation are adverse features. In resections, quantitation of the extent of residual viable tumor may be of greater clinical relevance than attempting to distinguish necrosis from other noncellular elements. We further demonstrate that the integration of clinical and microscopic features is a feasible and effective means for improving prognostication in patients with osteosarcoma. The effective stratification of patients using the proposed scoring system merits further investigation, ideally in prospective cohorts, to validate interobserver reproducibility and predictive ability. We thank Karen H. Wong (Deptartment of Statistics, University of Toronto) for assistance with statistical analysis, and Catherine T.- S. Chung, MD, for critical review of the manuscript. Only internal departmental funds were used for the work conducted in this study. References 1. Klein MJ, Siegal GP. Osteosarcoma: anatomic and histologic variants. Am J Clin Pathol. 2006;125(4): Ottaviani G, Jaffe N. The epidemiology of osteosarcoma. Cancer Treat Res. 2009;152: Fletcher CD, Bridge J, Hogendoorn PC, Mertens F, eds. The World Health Organization Classification of Tumours of Soft Tissue and Bone. Lyon, France: IARC; Zalupski MM, Rankin C, Ryan JR, et al. Adjuvant therapy of osteosarcoma a phase II trial: Southwest Oncology Group study Cancer. 2004;100(4): Scranton PE Jr, DeCicco FA, Totten RS, Yunis EJ. Prognostic factors in osteosarcoma: a review of 20 year s experience at the University of Pittsburgh Health Center Hospitals. Cancer. 1975;36(6): Taylor WF, Ivins JC, Unni KK, Beabout JW, Golenzer HJ, Black LE. Prognostic variables in osteosarcoma: a multi-institutional study. J Natl Cancer Inst. 1989;81(1): Bielack SS, Kempf-Bielack B, Delling G, et al. Prognostic factors in highgrade osteosarcoma of the extremities or trunk: an analysis of 1,702 patients treated on neoadjuvant cooperative osteosarcoma study group protocols. J Clin Oncol. 2002;20(3): Whelan JS, Jinks RC, McTiernan A, et al. Survival from high-grade localised extremity osteosarcoma: combined results and prognostic factors from three European Osteosarcoma Intergroup randomised controlled trials. Ann Oncol. 2012;23(6): Xing D, Qasem SA, Owusu K, Zhang K, Siegal GP, Wei S. Changing prognostic factors in osteosarcoma: analysis of 381 cases from two institutions. Hum Pathol. 2014;45(8): Meister P, Konrad E, Lob G, Janka G, Keyl W, Sturz H. Osteosarcoma: histological evaluation and grading. Arch Orthop Trauma Surg. 1979;94(2): Ohno T, Abe M, Tateishi A, Kako K, Miki H. Osteogenic sarcoma: a study of one hundred and thirty cases. J Bone Joint Surg Am. 1975;57(3): Hauben EI, Weeden S, Pringle J, Van Marck EA, Hogendoorn PC. Does the histological subtype of high-grade central osteosarcoma influence the response to treatment with chemotherapy and does it affect overall survival?: a study on 570 patients of two consecutive trials of the European Osteosarcoma Intergroup. Eur J Cancer. 2002;38(9): Bentzen SM, Poulsen HS, Kaae S, et al. Prognostic factors in osteosarcomas: a regression analysis. Cancer. 1988;62(1): Bacci G, Longhi A, Versari M, Mercuri M, Briccoli A, Picci P. Prognostic factors for osteosarcoma of the extremity treated with neoadjuvant chemotherapy: 15-year experience in 789 patients treated at a single institution. Cancer. 2006;106(5): Rosen G, Caparros B, Huvos AG et al. Preoperative chemotherapy for osteogenic sarcoma: selection of postoperative adjuvant chemotherapy based on the response of the primary tumor to preoperative chemotherapy. Cancer. 1982; 49(6): Glasser DB, Lane JM, Huvos AG, Marcove RC, Rosen G. Survival, prognosis, and therapeutic response in osteogenic sarcoma: the Memorial Hospital experience. Cancer. 1992;69(3): Lewis IJ, Nooij MA, Whelan J, et al. Improvement in histologic response but not survival in osteosarcoma patients treated with intensified chemotherapy: a randomized phase III trial of the European Osteosarcoma Intergroup. J Natl Cancer Inst. 2007;99(2): Davis AM, Bell RS, Goodwin PJ. Prognostic factors in osteosarcoma: a critical review. J Clin Oncol. 1994;12(2): McTiernan A, Jinks RC, Sydes MR, et al. Presence of chemotherapyinduced toxicity predicts improved survival in patients with localised extremity osteosarcoma treated with doxorubicin and cisplatin: a report from the European Osteosarcoma Intergroup. Eur J Cancer. 2012;48(5): Bloom HJ, Richardson WW. Histological grading and prognosis in breast cancer; a study of 1409 cases of which 359 have been followed for 15 years. Br J Cancer. 1957;11(3): Song WS, Kong CB, Jeon DG, et al. Prognosis of extremity osteosarcoma in patients aged years: a cohort/case controlled study at a single institute. Eur J Surg Oncol. 2010;36(5): Edge SB, Byrd DR, Compton CC, Fritz AG, Greene FL, Trotti A, eds. AJCC Cancer Staging Manual. 7th ed. New York, NY: Springer; Trojani M, Contesso G, Coindre JM, et al. Soft-tissue sarcomas of adults; study of pathological prognostic variables and definition of a histopathological grading system. Int J Cancer. 1984;33(1): Gustafson P. Soft tissue sarcoma: epidemiology and prognosis in 508 patients. Acta Orthop Scand Suppl. 1994;259: van Unnik JA, Coindre JM, Contesso C et al. Grading of soft tissue sarcomas: experience of the EORTC Soft Tissue and Bone Sarcoma Group. Eur J Cancer. 1993;15(93): Salzer-Kuntschik M, Brand G, Delling G. Determination of the degree of morphological regression following chemotherapy in malignant bone tumors [in German]. Pathologe. 1983;4(3): Min HS, Kang HG, Ro JY. Therapy related changes in osteosarcoma and ewing sarcoma of bone. Open Pathol J. 2009;3: Lucas DR, Kshirsagar MP, Biermann JS, et al. Histologic alterations from neoadjuvant chemotherapy in high-grade extremity soft tissue sarcoma: clinicopathological correlation. Oncologist. 2008;13(4): Coffin CM, Lowichik A, Zhou H. Treatment effects in pediatric soft tissue and bone tumors: practical considerations for the pathologist. Am J Clin Pathol. 2005;123(1): Bjornsson J, Inwards CY, Wold LE, Sim FH, Taylor WF. Prognostic significance of spontaneous tumour necrosis in osteosarcoma. Virchows Arch A Pathol Anat Histopathol. 1993;423(3): Harting MT, Lally KP, Andrassy RJ, et al. Age as a prognostic factor for patients with osteosarcoma: an analysis of 438 patients. J Cancer Res Clin Oncol. 2010;136(4): Pakos EE, Nearchou AD, Grimer RJ, et al. Prognostic factors and outcomes for osteosarcoma: an international collaboration. Eur J Cancer. 2009;45(13): Arch Pathol Lab Med Vol 140, November 2016 Prognostic Factors in Osteosarcoma Chui et al 1241

12 33. Bacci G, Bertoni F, Longhi A, et al. Neoadjuvant chemotherapy for highgrade central osteosarcoma of the extremity: histologic response to preoperative chemotherapy correlates with histologic subtype of the tumor. Cancer. 2003; 97(12): Dahlin DC, Unni KK. Osteosarcoma of bone and its important recognizable varieties. Am J Surg Pathol. 1977;1(1): Kraybill WG, Harris J, Spiro IJ, et al. Phase II study of neoadjuvant chemotherapy and radiation therapy in the management of high-risk, high-grade, soft tissue sarcomas of the extremities and body wall: Radiation Therapy Oncology Group Trial J Clin Oncol. 2006;24(4): Bacci G, Mercuri M, Longhi A, et al. Grade of chemotherapy-induced necrosis as a predictor of local and systemic control in 881 patients with nonmetastatic osteosarcoma of the extremities treated with neoadjuvant chemotherapy in a single institution. Eur J Cancer. 2005;41(14): Picci P, Sangiorgi L, Rougraff BT, Neff JR, Casadei R, Campanacci M. Relationship of chemotherapy-induced necrosis and surgical margins to local recurrence in osteosarcoma. J Clin Oncol. 1994;12(12): von Minckwitz G, Schmitt WD, Loibl S, et al. Ki67 measured after neoadjuvant chemotherapy for primary breast cancer. Clin Cancer Res. 2013; 19(16): Ogura K, Fujiwara T, Yasunaga H, et al. Development and external validation of nomograms predicting distant metastases and overall survival after neoadjuvant chemotherapy and surgery for patients with nonmetastatic osteosarcoma: a multi-institutional study. Cancer. 2015;121(21): Arch Pathol Lab Med Vol 140, November 2016 Prognostic Factors in Osteosarcoma Chui et al