Fax: (801) 585-3831
Concordance of histopathologic and cytologic grading in musculoskeletal sarcomas
Can grades obtained from analysis of the fine-needle aspirates serve as the basis for therapeutic decisions?
Article first published online: 2 APR 2002
Copyright © 2002 American Cancer Society
Volume 96, Issue 2, pages 83–91, 25 April 2002
How to Cite
Jones, C., Liu, K., Hirschowitz, S., Klipfel, N. and Layfield, L. J. (2002), Concordance of histopathologic and cytologic grading in musculoskeletal sarcomas. Cancer, 96: 83–91. doi: 10.1002/cncr.10479
- Issue published online: 8 APR 2002
- Article first published online: 2 APR 2002
- Manuscript Accepted: 12 OCT 2001
- Manuscript Revised: 17 SEP 2001
- Manuscript Received: 6 JUN 2001
- fine-needle aspiration;
- cytologic grade;
- histologic grade
Utilization of fine-needle aspiration (FNA) for the diagnosis of musculoskeletal lesions has been de-emphasized by many clinicians because of concerns regarding the technique's ability to supply all information necessary for prognostication and appropriate selection of therapy. Paramount among the concerns is the belief that FNA is incapable of supplying precise subtyping and grading in many cases. Secondary concerns regarding the use of FNA involve its perceived inability to supply adequate tissue for ancillary studies including immunohistochemistry and molecular diagnostic analysis. The authors investigated the ability of FNA to accurately subtype and grade a series of 107 primary sarcomas of the musculoskeletal system.
The files of the Departments of Pathology at Duke University Medical Center and the University of California at Los Angeles and the private consultation files of one of the authors were searched for all fine-needle aspirates of sarcomas arising within the musculoskeletal system. A total of 107 cases were obtained and reviewed by three board-certified cytopathologists. Each cytopathologist independently assigned subtype when possible and generated a grade for each of the sarcomas. Corresponding surgical material was available for 77 cases. The surgical material was re-reviewed for accuracy of diagnosis and assignment of grade independently of the cytologic examination. Correlation of cytologic grade with histopathologic grade was made and analyzed by the kappa test. In addition, agreement on grade between cytopathologists was analyzed, and accuracy of histologic subtype prediction by cytologic analysis was studied.
Most of the sarcomas were cytologically graded as Grade 1 or 2 but were assigned a histologic grade of 2 or 3. Cytopathologist A graded 93% of the sarcomas as either Grade 1 or 2, cytopathologist B graded 89% of cases as Grade 1 or 2, and cytopathologist C graded 94% of cases as Grade 1 or 2. Histologic evaluation demonstrated 82% of cases to be Grade 2 or 3. Correlation of cytologic and surgical grade varied among the observers. Cytopathologist A showed the highest degree of correlation between cytologic and surgical grade with an r value of 0.5. The corresponding r values for cytopathologists B and C were 0.46 and 0.41, respectively. Correlation between cytologic and surgical grade was significant for all three observers (P < 0.001). Only nuclear grade showed a consistent correlation in predicting final surgical pathology grade. Cellularity, mitotic rate, and the presence of necrosis were not statistically significant for predicting histopathologic grade. Accurate exact subtyping by cytologic examination was achieved in approximately 55% of cases. Prediction of histopathologic subtype by cytologic analysis was most successful when a distinctive stroma was present or high-grade features indicative of malignant fibrous histiocytoma were observed. There was little agreement as to histopathologic type as predicated by cytology for low-grade spindle cell sarcomas.
Although there was a statistically significant correlation between cytologically assigned grade and final histopathologic grade, statistical analysis revealed only a moderate correlation between the two with an overall r value of approximately 0.57. Cytologic analysis tended to undergrade in comparison to final histopathologic grading. Only analysis of nuclear atypia showed good correlation with final surgical grade. Fine-needle aspiration was only moderately successful at predicting histopathologic subtype for musculoskeletal sarcomas in this series. Cancer (Cancer Cytopathol) 2002. © 2002 American Cancer Society.
The utility of any diagnostic pathology report depends on its ability to provide prognostic data and all necessary information on which to base therapeutic decisions. The information required by the clinician from a pathologist's review of biopsy material depends on the type and site of the neoplasm being biopsied. Guidelines have been issued by the Association of Directors of Anatomic and Surgical Pathology for a variety of body sites, as well as by the College of American Pathologists.1, 2 The recommendations by the Association of Directors of Anatomic and Surgical Pathology1 suggests a diagnostic checklist for soft tissue sarcomas that includes the description of histologic type, histologic grade, and semiquantitation of the amount of necrosis present. These morphologic features appear to have both prognostic value as well as aid in the determination of the extent and type of therapy necessary. Although exact quantification of the extent of necrosis beyond its simple presence or absence in a needle aspiration specimen is not possible, documentation of histologic type and grade of the neoplasm should be achievable in most cases by cytologic examination.
Several studies have documented the diagnostic accuracy of fine-needle aspiration (FNA) in separating benign from malignant neoplasms of bone and soft tissue.3–11 Some of these studies have addressed the accuracy of FNA in histologic subtyping of mesenchymal tumors.3–11 The success of subtyping tumors of the musculoskeletal system has varied among studies but generally has been between 50% and 70% accurate.6–10 Some authors have demonstrated greater accuracy of subclassification for bone neoplasms,12 whereas others have shown significantly less success with subclassification of soft tissue neoplasms.4 Needle aspiration cytology is more successful at separating benign from malignant musculoskeletal lesions with diagnostic accuracy ranging between 80% and 95%.3–12 This rate of diagnostic accuracy compares favorably with that achievable by cutting needle biopsy.13 In the study by Mankin et al.,13 major diagnostic errors were found in up to 18.2% of surgical biopsies of musculoskeletal lesions. Seventeen percent of core needle biopsies were followed by significant technical complications significantly altering the optimal treatment plan.13 For reporting soft tissue sarcomas, the Association of Directors of Anatomic and Surgical Pathology recommend that sampling error is possible with needle biopsies, and therefore it is possible that needle biopsy is not an adequate procedure in making a diagnosis. Additional tissue may be needed, e.g., an incisional biopsy when appropriate. Most important, a needle biopsy should be used to establish the presence of malignancy and to determine whether a lesion is mesenchymal. From our data and that published within the literature,3–12 FNA appears to perform many of the same functions as core needle biopsy.
In addition to histologic typing of the neoplasm, the Association of Directors of Anatomic and Surgical Pathology recommend that a histologic grade be assigned to all sarcomas.1 Current therapy for adult soft tissue sarcomas and many sarcomas of bone is based primarily on anatomic location and stage.14, 15 In the staging systems currently utilized, tumor grade is an important component.14, 15 Thus, with current management techniques, grade appears to be more useful clinically than histologic subtype for the assignment of optimal therapy. Localized sarcomas are treated based on size and tumor grade. Localized sarcomas less than 5 cm of any grade initially are treated with wide excision. When margins are positive or close, external beam radiotherapy is added to the regimen. Low-grade sarcomas (Grade 1) larger than 5 cm also are treated initially with wide excision, and if positive or close margins are found at pathologic examination external beam radiotherapy is added to the protocol. Grade 2 and higher grade sarcomas greater than 5 cm in size initially are treated with neoadjuvant chemotherapy followed by wide excision. Hence, in this algorithm, grade rather than histologic subtype aids in the determination of optimal therapy.
Soft tissue sarcomas are relatively rare lesions with only approximately 4200 cases being accessioned into the National Cancer Data Base for 1993.15 This limited number of cases has inhibited the development of widely tested and accepted grading schemes for histopathologic analysis.16, 17 Although no single uniformly accepted grading scheme exists, the Association of Directors of Anatomic and Surgical Pathology recommends the utilization of either the French system or the National Cancer Institute grading protocol.1 The grading scheme used by the National Cancer Institute17 estimates the grade of malignancy by using a combination of the histologic subtype and a composition of histologic parameters that include cellularity, nuclear pleomorphism, degree of necrosis, and mitotic activity.17 The system proposed by Coindre et al.18 is based on a summation score derived from the individual scoring of tumor differentiation, mitotic count, and tumor necrosis. This results in a three grade system relatively independent of histopathologic subtype. Few published proposals for grading of soft tissue sarcomas by FNA exist. The system proposed by Kilpatrick et al.19 is based largely on tumor subtype or cytomorphologic group. The weakness of this system is that it is reliant on the cytopathologist's ability to subtype sarcomas, a process shown to be only 60–70% accurate in most studies. The proposal by Weir et al.20 uses a three grade system based on assessment of nuclear atypia, nuclear overlap, mitotic figures, and necrosis. This scheme is less dependent on tumor subtype. We investigated intraobserver variability of musculoskeletal sarcoma grading by FNA and its correlation with histologic grade in a series of more than 100 musculoskeletal sarcomas.
MATERIALS AND METHODS
The files of the Departments of Pathology at Duke University Medical Center and the University of California at Los Angeles, along with the private consultation files of one of the authors (L.J.L.) were searched for all cases of fine-needle aspirates performed from 1992 through 1999 with a diagnosis of sarcoma of the musculoskeletal system. Ewing sarcomas were excluded. A total of 107 cases were retrieved for which adequate cytologic material and follow-up was available. The histologic types of the sarcomas undergoing grading by cytologic means are shown in Table 1. Figure 1A–C show the degree of nuclear atypia for the corresponding scores 1, 2, and 3. Figure 2A–C show the degree of cellularity for scores 1, 2, and 3. Corresponding surgical material was available in 77 of these cases. These cases were blinded as to patient identifiers and sent to three cytopathologists for grading. The grading was performed independent of any knowledge of prior diagnosis, grade originally assigned to the cytologic and/or surgical specimen, and without knowledge of grade assigned by the other reviewers. Each neoplasm was cytologically scored using the system shown in Table 2. The cytopathologists (C.J., L.J.L., K.L.) all had prior experience with the FNA diagnosis of sarcomas. Grading was performed utilizing both Papanicolaou-stained preparations and air-dried preparations stained with a modified Romanowsky technique. Independent scores for cellularity, nuclear atypia, mitotic rate, and necrosis were determined. The final grade was calculated from the summation of these four individual scores, with Grade 1 assigned to those neoplasms with a summation score between 4 and 6, Grade 2 to those neoplasms having a summation score of 7 to 9, and Grade 3 to those neoplasms having a summation score of 10 to 12.
|Surgical pathology diagnosis||No. of cases|
|Malignant fibrous histiocytoma||24|
|Malignant peripheral nerve sheath tumor||2|
|Sarcoma not otherwise specified||14|
|Features scored||Points assigned|
|Cellularity||1 (low), 2 (moderate), 3 (high)|
|Nuclear atypia||1 (minimal), 2 (moderate), 3 (marked)|
|No. of mitotic figures/200 cells||1 (0–2), 2 (3–5), 3 (> 5)|
|Tumor necrosis||1, absent; 3, present|
After cytologic grading, the corresponding surgical biopsies and/or resection specimens were independently graded by one of the authors (L.J.L.) without knowledge of cytologic grade or histologic grade assigned at the time of original diagnosis. Histologic grade was based on the system described by Tsujimoto et al.21
Interobserver agreement for smear-based grades was determined for the individual characteristics of cellularity, nuclear atypia, mitotic rate, and necrosis, as well as for overall grade. The agreement was analyzed statistically using the kappa score technique. Correlation between individual observer's cytologic grades and final surgical grade was performed using r values. This analysis was conducted for both the individual cytomorphologic features of cellularity, nuclear atypia, mitotic rate, necrosis, and final overall grade. The percentage of cases correctly graded by each observer also was calculated.
Each observer attempted to assign a tumor subtype to all 107 cytology cases, and this subtype was correlated with the histologic subtype assigned at separate histopathologic review.
Table 3 catalogs the grade sum score distribution for the cytologic analysis by observer. This table also records the distribution of surgical grades by summation score. Table 4 documents the final grade distributions for cytologic analysis by each of the three observers and the distribution of surgical pathology grade. Most cases were graded as Grades 1 or 2 by cytologic examination, but as Grade 2 or 3 by histologic examination. Cytopathologist A graded 99 cases (93%) as either Grade 1 or 2. Cytopathologist B graded 95 cases (89%) as Grade 1 or 2, and cytopathologist C graded 100 cases (94%) as Grade 1 or 2. Histologic evaluation demonstrated that 63 of 77 (82%) cases undergoing histologic grading were Grade 2 or 3. Correlation of cytologic and surgical grade varied among the observers. Cytopathologist A correctly graded 32 of 75 cases (43%; r = 0.50). Cytopathologist B correctly graded 37 of 74 cases (50%; r = 0.46), and cytopathologist C correctly graded 33 of 74 cases (45%; r = 0.41). Correlation between cytologic and surgical pathologic grade was significant for all three observers, with a P value of less than 0.001. The r value for overall agreement between cytologic grade and histopathologic grade was 0.57. Of the cytologic variables, only nuclear atypia showed a consistent correlation in predicting final surgical pathology grade. The r value for correlation of nuclear atypia with final surgical pathology grade was best for cytopathologist C with an r value of 0.56. The r value correlation was 0.54 for cytopathologist A and 0.41 for cytopathologist B. The predictive value of nuclear atypia for final surgical pathology grade was statistically significant for each observer. The other cytologic variables of cellularity, mitotic rate, and necrosis were not significant (Table 5).
|Score sum||Observer A||Observer B||Observer C||Surgical scores|
|Grade||Observer A||Observer B||Observer C||Surgical grade|
|Features scored||Betas||Raw r values|
|r = 0.57, r2 = 0.32, F(4, 70) = 8.35, P < 0.001|
|Cellularity||−0.02 NS||0.11 NS|
|Nuclear atypiaa||0.54 P < 0.001||0.56 P < 0.001|
|Mitoses||0.02 NS||0.11 NS|
|Necrosis||0.10 NS||0.20 P < 0.08|
|r = 0.50, r2 = 0.25, F(4, 69) = 5.70, P < 0.001|
|Cellularity||−0.01 NS||0.15 NS|
|Nuclear atypia||0.41 P < 0.002||0.49 P < 0.001|
|Mitoses||0.05 NS||0.18 NS|
|Necrosis||0.11 NS||0.35 P < 0.001|
|r = 0.58, r2 = 0.33, F(4, 69) = 8.65, P < 0.001|
|Cellularity||−0.14 NS||0.13 NS|
|Nuclear atypia||0.56 P < 0.001||0.56 P < 0.001|
|Mitoses||0.02 NS||0.15 NS|
|Necrosis||0.02 NS||0.05 NS|
Examination of the data revealed that on average each rater underestimated surgical grade using cytologic criteria for prediction. Observer A correctly estimated final surgical pathology grade in only 42.7% of cases. Observer B correctly estimated grade in 50% of cases, and observer C correctly predicted final surgical grade by cytologic analysis in only 44.6% of cases. When sarcomas were divided into low- and high-grade categories, we still undergraded a significant percentage of cases (Table 6).
|Observer||Cytologic grade||Surgical grade||Statistics|
|A||Low||14||30||χ2(1) = 12.13, P < 0.001, n = 75|
|B||Low||12||24||χ2(1) = 12.03, P < 0.001, n = 74|
|C||Low||11||28||χ2(1) = 4.64, P < 0.4, n = 74|
A specific tumor subtype was assigned cytologically by reviewer A in 75% of cases, by reviewer B in 79% of cases, and by reviewer C in 73% of cases. This subtype agreed exactly with the final subtype assigned histologically in 58%, 52%, and 54% of cases, respectively. In 14 cases, an exact histologic subtype had not been assigned by surgical pathology review, and these cases were excluded from the analysis.
Fine-needle aspiration is a generally accepted technique for the workup of metastatic disease and primary neoplasms at a variety of locations.19, 22 However, it is less accepted as a diagnostic technique for the investigation of primary neoplasms of the musculoskeletal system. The guidelines recently published by the Association of Directors of Anatomic and Surgical Pathology have questioned the utility of the technique for the investigation of musculoskeletal lesions.1 These authors find the technique suboptimal because of concerns regarding precise typing and grading of these lesions and their belief that soft tissue tumors are susceptible to misdiagnosis by this technique because of morphologic heterogeneity.1 In addition to these objections raised by the Association of Directors of Anatomic and Surgical Pathology, others have suggested the FNA is not suitable for the investigation of these lesions because it supplies insufficient tissue to perform a variety of ancillary tests including immunohistochemistry and molecular diagnostic techniques. Although most authors admit that FNA is in general a cost-effective technique and some studies have shown that its accuracy rate closely parallels that of cutting needle biopsy,8, 10 FNA has not achieved widespread acceptance as the initial primary diagnostic technique for obtaining a tissue diagnosis in these lesions.
The diagnostic accuracy of FNA has been shown to compare favorably with that of core needle biopsy.7 In an initial study, Layfield et al.7 demonstrated that FNA was no more likely to result in a major diagnostic error or to complicate subsequent therapy than was core biopsy. The predominate issues in selecting a biopsy technique are whether the method chosen will supply all necessary prognostic and diagnostic information on which to base therapeutic decisions and whether the biopsy technique can avoid complications including unnecessary disruption of the tumor bed. Fine-needle aspiration is a minimally invasive technique and in general causes little if any disruption of the tumor bed favoring its use over incisional biopsy and in many cases core needle biopsy. The basic issues that need to be resolved in the comparison of FNA to other biopsy techniques is whether FNA can reliably supply sufficient information to allow selection of appropriate therapy and give pertinent prognostic information. Current therapy for adult soft tissue sarcomas is based predominantly on anatomic location, stage, and tumor grade.14 Current treatment protocols rely little on histologic subtype. The treatment protocol currently used at the University of Utah Medical Center relies only on size and tumor grade for the selection of therapy in localized soft tissue sarcomas (Fig. 1). Hence, the major issues confronting a clinician when deciding whether to use FNA for the workup of a sarcoma are whether the technique is reliable for separating benign from malignant lesions and whether the technique can reproducibly grade sarcomas. In the current study, we limited our investigation to non-small cell sarcomas. We studied the concordance of cytologic grade and histologic grade in a subset of 77 cases and investigated the interobserver reproducibility of cytologic grading in 107 cases. For the purposes of cytologic grading, we used a modified version of the scheme proposed by Weir et al.20 Although no universally accepted protocol exists for histologic grading, we followed the recommendations of Tsujimoto et al.21 In a study of 36 FNA specimens, Weir et al.20 accurately graded 92% of them using a 3-tier grading system. To achieve this degree of accuracy, the system was modified to rely more heavily on nuclear atypia whenever mitotic activity and/or necrosis were lacking.20 In our larger series of sarcomas, we were less successful in predicting final histopathologic grade by review of FNA specimens. The three observers in our study accurately predicted histopathologic grade in between 43% and 50% of cases. In 46% of cases, 2 of 3 or 3 of 3 observers accurately estimated final histologic grade by cytologic analysis. The cytologic grading system proposed by Kilpatrick et al.6 relies to a considerable extent on accurate cytologic subtyping of the neoplasms. In our current study, accurate cytologic subtyping occurred in between 52% and 58% of cases (50–70% in prior studies).7–10 In the study by Kilpatrick et al.19 only 54% of soft tissue sarcomas were accurately subtyped. From these data, it appears that FNA can give an accurate grade using the method of Kilpatrick et al.19 in approximately one-half of cases.
Our data indicate that cytologic grading tends to underestimate final histologic grade in most cases. In our study, cytopathologist A undergraded the sarcoma in 55% of cases and failed to separate low-grade (Grade 1) from high-grade (Grades 2 and 3) sarcoma in 40% of cases. Cytopathologist B undergraded 43% of cases and failed to distinguish high-grade from low-grade neoplasms in 32% of cases. Cytopathologist C undergraded half of all cases and failed to separate high-grade from low-grade sarcomas in 38% of cases. Hence, in approximately one-third of cases the cytologic grade would have been inappropriately low and resulted in undertreatment of patients if it had been relied on as the basis for selection of therapy using a protocol similar to that in effect at the University of Utah (Fig. 3). Palmer et al.23 achieved a 90% agreement between cytologic and histologic grade when a 2-grade system (low and high) was used. This dichotomization reflects clinical practice. We were unable achieve a similar accuracy when separating sarcomas into low (Grade 1) and high (Grade 2, 3, and 4) categories.
In our study, only nuclear atypia reached statistical significance for cytologic prediction of final histologic grade. Smear cellularity, mitotic rate, and the presence of necrosis were not predictive variables. Necrosis was so infrequent in our material that it did not occur in smears from even a majority of high-grade sarcomas. When comparing mitotic rates in smears and histologic sections of the same sarcomas, the smear preparations tended to contain far fewer mitotic figures resulting in an underestimate of the mitotic activity observed histologically. The reason that cellularity was not a predictor of grade in cytologic smears was unclear, but there did not appear to be a close correlation between smear cellularity and the density of cells in histologic sections.
In the current study, subclassification of musculoskeletal sarcomas by cytologic evaluation was accurate in 55% of cases. This is similar to the subclassification accuracy of 50–70% achieved in most previously published studies.6–11 Palmer et al.23 were only able to successfully subtype 14% of their cases. In our hands, FNA was most accurate in subtyping sarcomas when they possessed a distinctive stromal component. Palmer et al.23 found myxoid and spindle cell lesions most difficult to classify. In our study, most high-grade sarcomas were malignant fibrous histiocytomas, and these were also most likely to be accurately identified by cytologic analysis. Our success in identifying malignant fibrous histiocytoma differs from that of Bernardo et al.24 but may reflect the mixture of sarcoma subtypes in the current study or the study design.
Kilpatrick et al.19 have demonstrated that cytogenetic analysis is possible and accurate using material obtained by FNA. Cytogenetic analysis of aspirated material was able to confirm t(11;22) in two Ewing sarcomas and t(X;18) in three synovial sarcomas. The development of further cytogenetic analyses and molecular diagnostic techniques may improve the diagnostic accuracy and utility of FNA cytology for the investigation of both soft tissue and skeletal sarcomas. In Kilpatrick et al.'s study,19 FNA material was sufficient for the initiation of definitive therapy in 87% of patients with sarcomas of bone and 83% of patients with soft tissue sarcomas. In their study, even though histologic subtyping of adult soft tissue sarcomas was sometimes impossible, this inability did not influence initial therapy in most cases. In contrast with adult soft tissue sarcomas, subtyping of pediatric sarcomas by FNA appears highly accurate with immunohistochemical and molecular diagnostic techniques supplying further prognostic and diagnostic information.19, 25, 26
In summary, our findings support the opinion19 that FNA can accurately subtype and grade musculoskeletal sarcoma in most cases. Unfortunately, cytologic analysis appears to undergrade a significant minority of cases reducing the overall utility of the technique when a low-grade sarcoma is diagnosed cytologically. In addition, recent histologic studies suggest that individual grading of main histologic types of adult soft tissue sarcoma is most accurate in predicting the occurrence of metastasis.27 Despite these limitations, the technique is valuable as the initial diagnostic investigation for these neoplasms, especially if a sarcoma is not the leading clinical diagnosis. In this circumstance, FNA can correctly identify the lesion as sarcomatous and give significant information on type and grade in most cases.
- 2Reporting on cancer specimens. Protocols and case summaries. Northfield, IL: College of American Pathologists; 1998., , , .
- 11Fine-needle aspiration biopsy of primary bone tumors. Clin Orthop. 2000; 373: 80–87., .
- 13The hazards of biopsy in patients with malignant primary bone lesions and soft tissue tumors. J Bones Joint Surg. 1982; 64A: 1121–1127., , .
- 24Fine-needle aspiration cytopathology of malignant fibrous histiocytoma. Cancer (Cancer Cytopath). 1997;81: 228–237., , , , .
- 27Predictive value of grade for metastasis development in the main histologic types of adult soft tissue sarcomas. A study of 1240 patients from the French Federation of Cancer Centers Sarcoma Group. Cancer. 2001;91: 1914–1926., , , et al.