Myxoinflammatory Fibroblastic Sarcoma

Abstract

In this study, we examine the clinicopathologic features of 104 cases of myxoinflammatory fibroblastic sarcoma (MIFS), a low-grade, inflammatory fibromyxoid tumor with a predilection to distal extremity soft tissue, and attempt to identify factors predictive of aggressive behavior. The study cohort consisted of 49 male and 55 female patients ranging in age from 17 to 83 (mean, 42; median, 39) years. The tumor arose primarily on the dorsal aspect of the distal extremities as a solitary and usually painless mass. Tumors ranged in size from 0.5 to 15 (mean, 3.2; median; 2.4) cm. Microscopically, tumors consisted of variably cellular and inflamed fibromyxoid tissue growing as a lobulated mass or as multiple nodules within subcutaneous tissue or along tendinofascial planes. Tumor cells ranged from plump spindled to more epithelioid cells with enlarged, vesicular nuclei. Characteristic of the process was a strikingly bizarre cell with an inclusion body–like nucleolus (85% of cases) and/or a smudgy hyperchromatic nucleus (51%) present in all but 7 cases. The mitotic rate per 50 high-power field ranged from 0 to 13 (mean, 2,9; median, 2) mitoses. Twenty-two tumors demonstrated 1 or more of the following atypical features: (1) foci with complex sarcoma-like vasculature; (2) hypercellular areas; and (3) increased mitotic activity or atypical mitotic figures. Immunohistochemically, tumor cells demonstrated immunoreactivity for vimentin (100%), D2-40 (86%), CD34 (50%), keratin(s) (33%), CD68 (27%), actin(s) (26%), desmin (9%), S-100 protein (7%), and epithelial membrane antigen (6%). Thirty of 59 patients (51%) with follow-up data suffered (at least) 1 local recurrence, and 1 patient developed metastatic disease after multiple local recurrences. Completeness of initial surgical excision was the only clinicopathologic parameter that statistically correlated with a lower incidence of recurrence (P = 0.004). Histologically atypical MIFS recurred more often than conventional tumors (67% vs. 47%), but the difference was not statistically significant (P = 0.35). Our study shows that histologic features often associated with more aggressive sarcomas do not substantially impact the morbidity of MIFS, and complete surgical excision provides the best chance for disease-free survival.

In 1998, 3 independent groups published studies delineating a peculiar low-grade mesenchymal tumor predilecting acral sites and exhibiting a high recurrence rate.^1–3 Microscopically, the process was characterized by the presence of strikingly bizarre ganglion-like tumor cells scattered amongst spindled and epithelioid cells within an inflamed myxoid and fibrosclerotic stroma. The relative uncertainty regarding the tumor’s histogenesis and differences in outcome data resulted in the use of descriptive terms for the process, including “inflammatory myxohyaline tumor of the distal extremities with virocyte or Reed-Sternberg-like cells,”¹ “acral myxoinflammatory fibroblastic sarcoma,”² and “inflammatory myxoid tumor of soft parts with bizarre giant cells.”³ Subsequently, the diagnostic appellation “myxoinflammatory fibroblastic sarcoma” (MIFS) was adopted by the World Health Organization,⁴ as some examples of the tumor were found in nonacral sites or were reported to metastasize.

Since its initial description, investigators have raised the possibility of a link between MIFS and other distinct mesenchymal tumors. Tumors showing histologic overlap between MIFS and another low-grade mesenchymal process, the hemosiderotic fibrolipomatous tumor (HFLT),^5,6 have been documented, and a t(1:10) and amplification of 3p12.13, both of which were first detected in MIFS,^7,8 have been subsequently identified in HFLT^6,9,10 and in tumors showing hybrid morphology,^5,6 suggesting a possible histogenetic relationship. In addition, purported examples of myxofibrosarcoma recurring as MIFS,¹¹ tumors with histologic features of MIFS and high-grade pleomorphic sarcoma,¹⁰ or metastatic MIFS^2,12,13 suggest histologic overlap between MIFS and other more aggressive sarcomas and encourage a search for clinicopathologic factors that might predict clinical behavior of MIFS and/or help separate the lesion from more aggressive mimics.

This study analyzes the clinicopathologic and immunohistochemical features of 104 examples of MIFS retrieved from the files of the Armed Forces Institute of Pathology/Joint Pathology Center. Our main purpose is to identify clinicopathologic parameters that correlate with recurrent behavior and determine the impact that focal hypercellularity, a complex sarcoma-like vascular component, and abnormal mitotic activity have on the clinical course of the tumor.

MATERIALS AND METHODS

Mesenchymal tumors from extremity soft tissue sites accessioned to the Armed Forces Institute of Pathology/Joint Pathology Center between 1961 and 2004 and coded as atypical fibrohistiocytic proliferation, myxoid malignant fibrous histiocytoma, pseudomyxoid malignant fibrous histiocytoma, inflammatory myxohyaline tumor, acral myxoinflammatory tumor, and MIFS were retrieved. All clinical data accompanying the case and histologic slides were reviewed. A total of 104 tumors with salient histologic features delineated for MIFS^1–3: namely, (1) a fibrosclerotic stroma punctuated by myxoid areas; (2) presence of atypical epithelioid (ganglion-like) mesenchymal cells with voluminous deeply eosinophilic cytoplasm and 1 or more irregularly contoured and clefted nuclei harboring a greatly enlarged nucleolus (viral inclusion body–like or “owl’s eye” nucleolus) and/or exhibiting smudgy heterochromasia; (3) a histiocyte-like tumor cell with cytoplasm distended by mucin-filled vacuoles (pseudolipoblast) and exhibiting varying degrees of nuclear atypia; and (3) an intralesional inflammatory component, were included in the study. Seventeen cohort cases were accessioned after the entity was initially described in 1998. Exclusionary criteria included diffuse and severe cytologic atypia, an exclusively myxoid stromal matrix, an extensive vascular network of branching, anastomosing, or thick-walled curvilinear vessels, or an exceedingly high mitotic count arbitrarily set at >20 mitoses per 50 high-power field (HPF). The following histologic parameters were recorded for each case: (1) location of the tumor; (2) proportion of myxoid to solid areas; (3) proportion of epithelioid to spindled tumor cells; (4) presence of characteristic atypical ganglion-like tumor cells with inclusion body–like nucleoli and/or a smudgy chromatin pattern; (5) presence of pseudolipoblasts; (6) overall cellularity of each lesion, including hypercellular foci with solid, whorled, storiform, or fascicular growth; (7) type and amount of necrosis; (8) concentration, distribution, and types of inflammatory cells; (9) presence of cells demonstrating emperiopolesis; (10) presence of fibrin and stromal or intracellular deposits of hemosiderin; and (11) mitotic activity in 50 most cellular HPFs (estimated total area 10 mm²) and presence of atypical mitotic figures.

Patient follow-up data were collected in or around 2007 and then between 2010 and 2012. Initial follow-up data were gathered through communications with physicians, medical record departments, and with patients, whereas later follow-up data were retrieved mainly through communications with physicians, pathologists, and medical record departments. Long-term clinical follow-up interval is defined as a time span of at least 5 years and short-term clinical follow-up interval as a time interval of <5 years after clinical presentation or first surgical intervention.

Immunohistochemical Analysis and Scoring

Immunohistochemical staining analyses were performed on 45 tumors with formalin-fixed, paraffin-embedded material. As the vast majority (84%) of the immunohistochemical tests were performed at the former Armed Forces Institute of Pathology or at the contributing outside hospital over a time period spanning close to 30 years, the antibody clones, their sources and dilutions, and retrieval methods used varied and are not readily traceable.

Clusterin immunohistochemical testing (n = 18) was performed on the Dako Autostainer Plus using antibody clone 41D (BioCare Medical, Concord, CA) at a dilution of 1:50 with a heat-induced epitope-retrieval method, which involves placing slides in a citrate buffer (0.01mol/L citric acid at pH 6.0) overnight in a 601C oven. Antipodoplanin immunohistochemical testing (n = 14) was performed on the Leica Bond Max using antibody clone D2-40 (Dako, Carpinteria, CA) at a dilution of 1:50. For epitope retrieval, slides were placed in epitope-retrieval solution (ER1) for 20 minutes in a 100°C oven.

Immunohistochemical expression of each marker was scored semiquantitatively in the following manner: score 0—no cells positive; score 1—1% to 24% of cells positive; score 2—25% to 49% of cells positive; score 3—50% or more of cells positive.

Statistical Analysis

Dichotomous parameters were analyzed for their effect on recurrence-free survival of patients with MIFS using the Fisher (2-tailed) exact test. A P value of <0.05 was considered significant.

CLINICAL DATA

The study cohort consisted of 49 male and 55 female patients ranging in age from 17 to 83 (mean, 42; median, 39) years. All lesions were solitary and arose in the soft tissues of the hand (n = 24), finger (n = 21), foot (n = 17), wrist (n = 14), knee and lower leg (n = 9), elbow and forearm (n = 5), ankle (n = 4), leg, not otherwise specified (NOS) (n = 3), upper leg (n = 2), and upper arm, shoulder, and inguinal region (n = 1, each). The site of tumor was unknown for 2 cases. The tumor occurred in fingers, hands, and feet in 61% of cases and arose in dorsal soft tissue in 73% of cases involving distal acral sites. An asymptomatic mass or swelling was the most common initial patient complaint registered. Of the 18 individuals who presented with pain or tenderness, 6 of them experienced recent onset of pain in a preexisting asymtomatic (painless) lesion. Duration of the process before clinical presentation or first surgical intervention ranged from 2 weeks to “years,” with 60% of patients reporting signs or symptoms for 1 year or less. Prior injury to the tumor site was documented in 10 cases and included blunt trauma or contusion (n = 7), animal or insect bite (n = 2), and a puncture wound (n = 1). Two members of the study carried a diagnosis of diabetes mellitus. A preoperative clinical impression was available in 13 cases and included ganglion cyst (n = 7) and synovitis, lipoma, accessory ossicle (or ganglion cyst), infected cyst, calcified bursa, and giant cell tumor (n = 1, each).

GROSS DESCRIPTION

A partial or less often complete macroscopic description of the tumor was available for 74 cases. The tumors measured from 0.5 to 15 cm in maximum dimension, with mean and median diameters of 3.2 and 2.4 cm, respectively. In 12 cases, the specimen was received in multiple fragments. Most specimens had a firm to rubbery consistency with a mucoid cut surface and ranged in color from a light tan-gray to yellow-pink. Nineteen tumors demonstrated a lobulated contour or nodular configuration (Fig. 1). Cystification and focal necrosis were identified in 2 tumors each. Surgical margins were inked for microscopic assessment in 7 cases.

FIGURE 1.

Macroscopic appearance of MIFS: fragmented multinodular mass with a glistening surface.

MICROSCOPIC FINDINGS

The majority of tumors was centered in subcutaneous tissue and typically grew as either a single lobulated nodule or less often as multiple ill-defined tumor nodules along fibroconnective septa of subcutaneous fat, fascial planes, or tendon sheaths (Fig. 2A). Seven tumors also involved the dermis, and 2 were observed invading underlying skeletal muscle.

FIGURE 2.

Low-power histologic patterns of MIFS. A, Tumor nodules proliferate along fibrous septa within subcutaneous fat. B, Most tumors demonstrated an inflamed fibrosclerotic stromal matrix interrupted by small pools of mucin. C, Bulbous hyalinized fibrous septa with granulation tissue–like vessels partition mucinous areas in a minority of cases. D, Myxoedematous bulbous fronds of lesional tissue are often times associated with tenosynovial tissue.

Low-power magnification highlighted the rich stromal matrix of the tumor that consisted primarily of fibrosclerotic stroma interrupted by small mucin-rich pools (75% of cases) (Fig. 2B). Large myxoid zones partitioned by fibrosclerotic bands or bulbous hyalinized fibroconnective septa with granulation tissue–like vessels were observed mostly as a focal finding in a minority of cases (Fig. 2C). The presence of bulbous fronds of myxoedematous lesional tissue was a focal or predominant growth pattern in 10 tumors (Fig. 2D).

The degree of cellularity and cellular composition varied considerably from lesion to lesion. Moderate to high cellularity was appreciated in 54 cases, whereas the cellularity of the remaining 50 lesions was relatively low. The lesional elements morphologically ranged from plump spindled cells with fibrillary cytoplasm to more epithelioid, histiocyte-like cells having variably enlarged, vesicular nuclei with marginated chromatin and distinct eosinophilic nucleoli (Fig. 3A). Seventy-five tumors were epithelioid cell predominant, whereas spindled cells were the main lesional element in the remaining 29 cases. Intranuclear cytoplasmic inclusions were noted in a significant percentage of cases. The hallmark of the process was the presence of epithelioid tumor cells with abundant eosinophilic cytoplasm identified in varying proportions in each case and exhibiting an enlarged nucleus (typically conspicuous at ×10 magnification) featuring (1) a viral inclusion body–like nucleolus (n = 88) (Fig. 3B) and/or (2) smudgy heterochromatin (n = 53) (Fig. 3C). In addition, pseudolipoblasts, some with enlarged nuclei and nucleoli (Fig. 3D), were identified in 65 cases. Twelve tumors showed either an absence of the characteristic atypical ganglion-like cell (n = 7), or the latter were so heavily masked by the inflammatory infiltrate (n = 5) that the lesion resembled an infectious or non-neoplastic inflammatory process (Fig. 4A). The mitotic rate per 50HPF in the neoplastic component ranged from 0 to 13 (mean, 2.9; median, 2) mitoses. Sixty tumors had 1 to 5 mitotic figures (59%), 12 tumors (12%) had 6 to 10 mitotic figures, 6 tumors (6%) contained > 10 mitoses, and no mitotic figures were detected in the remaining 24 lesions (24%) in 50 HPFs. One or more atypical mitotic figures were identified in 6 cases. Fifteen tumors had small microscopic foci of coagulative (n = 12) or fibrinoid (n = 3) necrosis.

FIGURE 3.

The cellular element of MIFS (A) ranges from tumor cells with mildly enlarged vesicular nuclei and conspicuous nucleoli (arrows) to large cells with abundant cytoplasm and nuclei with virocyte-like nucleoli. Tumor cells characteristic of the process include (B) ganglion-like cells with abundant cytoplasm and large vesicular nucleus with virocyte-like nucleolus, (C) epithelioid tumor cells with an enlarged, irregularly contoured nucleus with smudgy heterochromatin, and (D) atypical pseudolipoblast-like cells with cytoplasm markedly distended by mucin vacuoles.

FIGURE 4.

Less frequently encountered histologic findings in MIFS include (A) a marked inflammatory infiltrate masking the few and scattered signature tumor cells (arrow) simulating an inflammatory process, (B) large atypical tumor cells with a copious amount of cytoplasm exhibiting emperiopolesis of acute inflammatory cells, and (C) foci of mildly atypical spindled cells infiltrating subcutaneous fat with small deposits of hemosiderin (arrow) resembling the growth pattern of HFLT.

The number and types of inflammatory cells varied greatly from case to case. All tumors possessed a lymphoplasmacytic infiltrate, which was heavy in 15 cases and organized into follicles or aggregates or oriented around vessels generally at the periphery of the process; in most examples, it was also observed percolating through lesional tissue. Along with a lymphoplasmacytic component, neutrophils were observed in 72 cases, forming microabscesses in 9 examples, and eosinophils were identified in 52 tumors and were relatively prominent in 4 cases. Histiocytes were encountered in most cases and included scattered multinucleated forms, often times resembling osteoclasts (n = 22), clusters or sheets of xanthomatous histiocytes (n = 21), and Touton-like giant cells (n = 14). Emperiopolesis of inflammatory cells, mostly neutrophils, by scattered tumor cells was observed in 10 cases (Fig. 4B). Hemosiderin and/or fibrin deposits were noted in 63 cases, with 4 lesions showing marked hemosiderin deposition.

Four tumors with features of conventional MIFS focally demonstrated a proliferation of mildly atypical spindle cells accompanied by small deposits of hemosiderin within subcutaneous fat as observed in the HFLT^14,15 (Fig. 4C). Three tumors focally exhibited morphologic features reminiscent of the pleomorphic hyalinizing angiectatic tumor (PHAT),¹⁶ including small foci of ectatic and thrombosed vessels surrounded by spindled lesional cells in 2 cases and a single microscopic cellular nodule of atypical epithelioid cells with numerous intranuclear cytoplasmic inclusions (but no characteristic vascular changes) in the third case.

CLINICOPATHOLOGIC DATA OF MIFS CASES WITH ATYPICAL HISTOLOGIC FEATURES

Twenty-two tumors with histologic features of MIFS as described above also demonstrated (1) foci with complex sarcoma-like vasculature manifesting as branching and/or thick-walled arcuate vessels and involving between 20% to 75% of the tumor (n = 5) (Fig. 5A), (2) small areas in which spindled or epithelioid cells were arranged in cellular solid, whorled, fascicular, or storiform patterns (n = 4) (Fig. 5B), or (3) abnormal mitotic activity defined as > 10 mitoses per 50HPF and/or the presence of atypical mitotic figure (n = 4). An additional 5 tumors exhibited both foci of increased cellularity and abnormal mitotic activity (including the one tumor with a single hypercellular nodule containing cells resembling those of PHAT), and 4 tumors demonstrated sarcoma-like vascularity with focal increased cellularity (n = 2), abnormal mitotic activity (n = 1), or both features (n = 1). These atypical MIFS variants occurred in 12 male and 10 female patients ranging in age from 15 to 77 (mean, 41; median, 34) years of age. The tumors arose primarily in the acral soft tissue, with 60% arising in fingers, hands, or feet, and ranged in size from 1.0 to 15.0 (median, 3.4) cm.

FIGURE 5.

MIFS with atypical histologic features. A, Area of tumor with concentrated, thick-wall arcuate vessels. B, Cellular focus with a vague fascicular growth pattern of plump tumor cells.

HISTOLOGY OF RECURRENT LESIONS

The histology of 8 recurrent tumors was reviewed and compared with the primary. Four recurrences from conventional tumors, including 1 tumor with total absence of the characteristic atypical lesional cell and 1 focally exhibiting HFLT-like features, showed an increase in cellularity and number of atypical cells. An atypical MIFS with focal sarcoma-like vasculature comprising approximately 20% of the tumor recurred as a tumor resembling myxofibrosarcoma with uniform cytologic atypia and a more widespread complex vascular network. The recurrences from the remaining 2 conventional MIFSs and 1 atypical variant with increased mitotic activity were morphologically similar to the primary.

IMMUNOHISTOCHEMICAL RESULTS

Immunohistochemical studies were carried out on 45 tumors, and the results are tabulated in Table 1. Vimentin expression was identified in tumor cells in all 18 tumors tested and was strong and diffuse in 16 cases. Twelve of the 14 tumors evaluated for D2-40 expression demonstrated variable positivity in scattered spindled tumor cells but mostly in pseudolipoblasts (including highly atypical pseudolipoblasts in 5 cases) (Fig. 6A). CD34 expression was found in 50% of tumors tested, with ≤25% spindled tumor cells positive in 5 cases and between 30% and 80% of spindled elements in an additional 2 examples (Fig. 6B). Tumor cells in 1 tumor with PHAT-like vascular changes were negative for CD34. Pan-keratin expression was found in a small number (≤25%) of the epithelioid tumor cells including some of the hallmark ganglion-like tumor cells in 10 of the 30 tumors evaluated (Fig. 6C). One tumor expressed keratins 7 and 18, and another expressed keratin 8, but all other keratins examined, including keratin CAM5.2 and keratins 5, 6, 5/6, and 19, were negative. A small number (≤25%) of spindled tumor cells were highlighted by actin immunoreagents (32% with smooth muscle actin and 13% with muscle-specific actin). Anti-CD68 marked mostly small histiocytes, dendritic cells, and a small number of tumor cells in 4 of the 15 tumors tested. S-100 protein was expressed in scattered lesional cells in 2 cases and dendritic fibroblasts in many of the 27 tumors tested. One of 11 tumors showed rare spindled cells positive for desmin. EMA was observed to be positive in scattered cells of 1 tumor but negative in the remaining 16 evaluated. Anticlusterin did not stain lesional cells in the 18 tumors evaluated.

TABLE 1.

Immunohistochemical Results for 45 Patients With MIFS

	Overall IHC Score for Case
Immunoreagent	0	+1	+2	+3	% Cases (+)
Vimentin	0	1	1	16	100
Antipodoplanin (D2-40)	2	5	3	4	86
CD34	7	5	1	1	50
Keratins	20	10	0	0	33
SMA	13	6	0	0	32
Anti-CD68 (KP1)	11	3	1	0	27
MSA	13	2	0	0	13
Desmin	10	1	0	0	9
S-100 protein	25	2	0	0	7
EMA	16	1	0	0	6
Clusterin	18	0	0	0	0

Score 0: no cells positive; score 1: 1% to 24% of cells positive; score 2: 25% to 49% of cells positive; score 3: 50% or more of cells positive.

% Cases (+) indicates percentage of cases expressing protein.

EMA indicates epithelial membrane antigen; IHC, immunohistochemical; MSA, muscle-specific actin; SMA, smooth muscle actin.

FIGURE 6.

Immunohistochemical findings in MIFS. A, Focal D2-40 positivity primarily in pseudolipoblasts. This antibody was positive in 86% of cases. B, CD34 membranous positivity was noted in spindled tumor cells in 50% of cases. C, Pan-keratin expression within tumor cells was a focal finding in 33% of tumors.

CLINICAL OUTCOMES FOR 59 MIFS PATIENTS WITH FOLLOW-UP DATA

Long-term (≥ 5 y) follow-up data were available for 41 of the 104 patients (40%) (range, 64 to 372 mo; median, 172 mo). Eighteen individuals (17%) had short-term (< 5 y) follow-up data collected mostly from information sent by the pathologist at the time of initial consultation or recurrence.

An additional 16 patients died with the status of their tumor unknown at the time of death, and 17 patients were alive at last contact (range, 145 to 410 mo; mean, 259 mo), but the status of their tumor also was unknown. The remaining 12 study members were lost to follow-up. Thirty of the 59 patients (51%) with long-term or short-term follow-up data suffered locally recurrent disease (range, 3 to 108 mo; median, 13 mo), and 1 patient with locally recurrent atypical MIFS of the ankle developed ipsilateral upper extremity and groin metastases. None of the patients developed distant metastases. Of the 41 individuals with long-term data, 26 have remained disease free for at least 5 years (63%).

CLINICAL OUTCOMES FOR 47 PATIENTS WITH CONVENTIONAL MIFS

Of the 82 patients with conventional MIFS, long-term clinical data spanning an interval from 63 to 394 (median, 204) months were available for 34 patients (41%), and short-term clinical data were available for an additional 13 individuals (range, 9 to 48 mo; median, 15 mo).

Twenty-two of the 34 patients with long-term follow-up information (65%) information have remained disease free after initial surgical intervention (range, 63 to 342 mo; median, 192 mo), which included a local excision, NOS (n = 16) and an excision or biopsy followed shortly by either an amputation (n = 2) or reexcision (n = 4). On the basis of the gross description of the submitted material, surgical procedures we designated as “local excision, NOS” were mostly marginal or intralesional (including “piecemeal) excisions, or the description of the procedure was not specific as to its extent or completeness. Surgical margin status of the excision was documented for only 2 patients, with margins involved by the tumor in 1 and uninvolved in the other. Three patients with short-term follow-up information were free of disease 9 to 43 months after initial excision of their tumor.

Local recurrence was reported for 22 of the 47 patients with either long-term or short-term follow-up data (47%). Time interval to first recurrence ranged from 3 to 108 (mean, 33; median, 14) months. Fifteen patients suffered 1 recurrence, 2 individuals had 2 recurrences, and 5 experienced 3 or more recurrences. Nineteen patients were initially treated with a local excision (NOS), 2 individuals had a reexcision shortly after biopsy, and the initial procedure was not documented for the remaining patient. Surgical margins were reported as positive in 3 patients, 2 of whom developed a recurrence 8 months (on needle aspiration biopsy) and 12 months, respectively, after the initial procedure, and the remaining individual has remained free of disease for 78 months. One individual with recorded negative surgical margins after excision has remained disease free for 78 months.

Of the group suffering recurrent disease, 10 individuals have been disease free for an interval at least 5 years after treatment for their last recurrence (range, 60 to 384mo; median, 186mo), which included local excision (n = 6), and the addition of radiation therapy for 3 of these patients, wide local excision (n = 1), and amputation (n = 3).

Three of 5 patients (60%) with tumors totally lacking hallmark atypical cells were rendered disease free for over 5 years after initial surgery, which included a local excision (NOS) (n = 2) and amputation (n = 1). The remaining 2 patients, both of whom experienced a single recurrence, have remained disease free for over 5 years after their last surgical intervention.

CLINICAL OUTCOMES FOR 12 PATIENTS WITH HISTOLOGICALLY ATYPICAL MIFS

Seven of the 22 patients had long-term follow-up data (range, 72 to 372 mo; mean, 189 mo), and 5 individuals had short-term data available (range, 8 to 30 mo; mean, 18 mo).

Four of the 7 patients with long-term follow-up (57%) information did not have a recurrence. Two of these individuals were managed with local excision (NOS), and the remaining 2 patients had a wide local excision, with 1 also receiving postoperative chemoradiation.

Eight of the 12 individuals in this group suffered local recurrence (67%), ranging from 3 to 30 (mean, 15) months after initial local excision (NOS). This cohort included all 5 patients with short-term follow-up. Four patients experienced multiple recurrences. One female with an atypical MIFS of the ankle featuring focal high cellularity and abnormal mitotic activity developed 2 local recurrences and was finally treated with a below-knee amputation. During the next 2 years, she suffered metastatic spread to soft tissues of the thigh and groin and received chemotherapy and radiation treatment. Three years after her last thigh recurrence, she was placed on imatinib mesylate (Gleevac) and has been disease free for 59 months. Two individuals with atypical MIFS involving the finger developed 3 or more recurrences but have been disease free for 372 and 252 months after an amputation and wide local excision, respectively. The remaining patient, who suffered 2 recurrences, has no follow-up information available.

EFFECT OF SURGICAL PROCEDURE AND RADIATION ON DISEASE-FREE SURVIVAL

Comparative analysis of the surgical impact on recurrence is summarized in Table 2. Forty-four patients with clinical follow-up data were initially treated with a local excision, NOS (without the surgical margin status) or a local or wide excision with microscopically or macroscopically involved surgical margins (group A), and 10 patients were initially managed with an excision in which negative surgical margins were documented or had a reexcision or amputation shortly after an initial biopsy or excision (group B). Twenty-seven group A patients (61%) and 1 group B member (10%) suffered at least 1 recurrence. The remaining 17 group A and 9 group B patients had no documented recurrence for a period of at least 5 years after initial treatment. The treatment rendered to group B patients resulted in a statistically significant lower recurrence rate compared with that documented for group A (P = 0.004). Twelve patients with a history of at least 1 recurrence eventually experienced a disease-free interval of over 5 years after the last surgical procedure. Six of these individuals were managed with a local reexcision (NOS), and 6 underwent wide excision (n = 2) or amputation (n = 4).

TABLE 2.

Statistical Analysis of Prognostic Factors for MIFS in 59 Patients With Follow-up Data

Clinicopathologic Variable	No. With Recurrent Disease	No. Without Recurrence For ≥ 5y	Recurrence Rate (%)	P
Size (cm)				0.23
< 3	17	11	61
≥ 3	7	10	41
Stroma*				1.0
Solid > myxoid	19	18	51
Myxoid > solid	9	8	53
Necrosis				1.0
Necrosis	5	5	50
No necrosis	25	21	50
Hallmark cells (all cases)				0.65
Present (n = 54)†	28	23	52
Absent (n = 5)	2	3	40
Hallmark cells (conventional MIFS)				1.0
Present (n = 42)†	20	19	48
Absent (n = 5)	2	3	40
Histologic features				0.35
Conventional (n = 47)†	22	22	47
Atypical (n = 12)	8	4	67
Atypical histologic features				0.61
≤ 1 atypical feature	26	26	50
≥ 2 atypical feature	3	1	75
Surgical procedure				0.004
Group A	27	17	61
Group B	1	9	10

P value of <0.05 is a statistically significant difference.

Group A—patients initially treated with local excision, NOS, or an excision with documented involved surgical margins; Group B—patients initially treated with margin-negative excision, wide local excision, or amputation or who underwent prompt reexcision of the tumor.

^*Tumors with a subjective equal mixture of myxoid and collagenous stroma are not included.

^†Includes 3 patients with no evidence of disease after short-term follow-up.

Exc. indicates excision; hallmark cells include atypical ganglion-like cells and atypical pseudolipoblasts.

Eight patients were given radiation therapy during the course of their disease. Two patients with conventional MIFS received radiation after initial local excision. One suffered a recurrence 9 years after combined treatment, whereas the other patient has been free of recurrent disease for 64 months. Four patients with recurrent disease (3 with conventional and 1 with an atypical MIFS) received radiation along with tumor reexcision after recurrence and remained free of disease for over 5 years. Two additional patients with atypical MIFS received radiation after recurrence. One of these patients has been free of additional disease for 12 months. The other patient who suffered multiple recurrences with eventual thigh and groin metastasis from an atypical ankle MIFS was given radiation twice after the tumor recurred in the thigh following a below-knee amputation.

STATISTICAL ANALYSIS OF PROGNOSTIC FACTORS

The analysis of specific clinicopathologic parameters in the 59 patients with follow-up data is summarized in Table 2. Factors associated with a higher rate of recurrence included: (1) a size ≥ 3 cm; (2) presence of atypical histologic features; and (3) incomplete surgical excision. Conventional MIFS (with hallmark cells) recurred slightly more often than tumors having a total absence of hallmark tumor cells (48% vs. 40%) but less often than tumors with atypical histologic features (67%), although the number of cases in the latter 2 categories was relatively small. We were unable to analyze atypical histologic parameters individually because of the limited number of cases in each category but found that tumors demonstrating at least 2 atypical histologic features recurred more often than conventional tumors or those with only 1 atypical histologic feature (75% vs. 50%). Of the clinicopathologic features analyzed, only the adequacy of the initial surgical attempt to completely remove the tumor showed a statistically significant lower rate of tumor recurrence when compared with patients who were initially managed with an incomplete surgical excision (10% vs. 61%) (P = 0.004).

DISCUSSION

In our clinicopathologic analysis of 104 examples of MIFS, which is the largest compilation of cases to date, we found that the only clinicopathologic factor significantly associated with a lower incidence of recurrence for this low-grade neoplasm was the adequacy of the initial surgical excision. In addition, the correlations between microscopic foci of hypercellularity or sarcoma-like vasculature, or modestly increased mitotic activity and/or presence of an occasional atypical mitotic figure and the incidence of recurrence within an otherwise conventional MISF were not statistically significant, although the number of such cases was relatively low.

The clinical features of MIFS in our study cohort compared favorably with previously reported series.^{1–3,10,12,13,17–21} Patients in our study presented primarily in the fourth through sixth decades of life (median, 39y) and showed a nearly equal sex distribution. Tumor arose primarily in distal appendicular soft tissue with fingers, hands, or feet accounting for 61% of cases. To our knowledge, we are the first to document that tumors involving distal acral sites show a strong predilection for dorsal soft tissue. As noted in prior studies,^{1,2,12,13,18–20,22,23} the process usually presents in a benign manner as an asymptomatic mass or swelling. We found, as did others,^12,19,20 that the majority of patients reported evidence of disease ≤1 year before clinical intervention. Injury to the tumor site has only been reported in a handful of cases,^2,24 but we document 10 of our patients with a history of injury to the affected site up to 4 years before presentation. The clinical features of the 22 tumors with atypical histologic features did not appreciably differ from those of patients with conventional MIFS.

Pathologically, we observed, like others,^{1,2,13,17,19,22,23} a tendency for the tumor to grow along the interlobular septa of subcutaneous fat or along fascial planes or tendon sheaths, with only a minority of tumors involving the dermis and even fewer invading skeletal muscle. Meis-Kindblom and Kindblom² described a peculiar growth pattern of bulbous myxoedematous fronds of lesional tissue when MIFS involves synovial tissues. We identified this growth pattern in 10 cases, which include 3 tumors with a documented association with a joint or tendon. Histologically, all tumors in our study demonstrated a combination of the salient cytomorphologic findings initially described for MIFS,^1–3 including a predominantly fibrosclerotic stroma punctuated by variably sized mucinous pools (75% of cases), a variable number of large epithelioid cells with abundant eosinophilic or markedly distended mucin-filled cytoplasm and irregularly contoured nuclei with prominent “owls eye” nucleoli and/or smudgy heterochromatin (93%), pseudolipoblasts (63%), and an intralesional inflammatory infiltrate in all cases.

We also reaffirmed that infrequently reported findings of a nonacral tumor location¹⁷ (observed in 5 cases in our study), emperiopolesis of inflammatory cells by lesional cells^2,3,12,25,26 (10% of cases), and small foci of necrosis^1,2,12,19,27 (14%) are bonafide features observed in MIFS.

Seven tumors in our study showed an absence of signature large atypical tumor cells. We admit, however, that a limited number of available slides for review in some cases may have inflated the number of such cases. Nevertheless, 2 lines of data support their inclusion as examples of MIFS. First, this small subset of patients showed a significant recurrence rate (40%), which was comparable to that of the remaining examples of conventional MIFS (48%); second, 1 tumor with no hallmark atypical cells in the initial excision showed features of conventional MIFS replete with atypical cells on recurrence. On the basis of these observations, we believe that in the right setting, the identification of modestly atypical spindled or epithelioid cells, or pseudolipoblasts, even in small numbers should prompt strong consideration for the diagnosis of MIFS and continued search for the hallmark atypical lesional cells.

Four tumors in our study with features otherwise characteristic of MIFS contained small areas that overlapped histologically with HFLT^14,15 (n = 4) or with PHAT¹⁶ (n = 3) and which could have potentially led to misdiagnosis on a limited biopsy. Tumors with shared morphologic features of MIFS and HFLT^5,6 (both of which are known to share the exact same chromosomal translocation^5–10) or PHAT and HFLT (with the latter considered by some as the early manifestation of PHAT)²⁸ are well documented in the literature. However, unlike these purported hybrid tumors, which occurred primarily in the ankle area, 5 of our 7 cases arose in the upper extremity. We prefer to classify these tumors in our study as examples of MIFS and argue that the confounding histologic features represent superimposed reactive change, as the soft tissues of wrists, fingers, hands, and ankles are susceptible to trauma and/or vas-cular compromise.^14,29,30 In addition, we find it tempting to speculate that many such hybrid tumors, particularly those with histologic features of both MIFS and HFLT, which are known to share a common translocation,^5–10 can be explained in this manner.

An intricate network of capillary or arcuate vessels, abnormal mitotic activity, and hypercellularity are histologic features not previously emphasized in large studies of MIFS, particularly with regard to their effect on outcome. For example, most studies document a low mitotic index (usually 5 or less mitoses per 50 HPF),^1,3,12,13,17 whereas a few mention the presence of atypical mitotic figures,¹⁷ accentuated vascularity particularly in the myxoid areas,^2,3,7 or hypercellularity.^{1,12,18–20,22} We found that the incidence of recurrence of MIFS roughly correlates with increasing number of atypical histologic features, although the difference did not reach statistical significance and the number of atypical cases was relatively small. In addition, the presence of these findings may engender a diagnosis of a potentially more aggressive sarcoma; namely, myxofibrosarcoma,³¹ which also involves appendicular soft tissue and can show overlapping histology. In contrast to MIFS, myxofibrosarcoma typically arises in more proximal soft tissue sites and, microscopically, is a predominantly myxoid neoplasm that demonstrates a more extensive network of arcuate and branching vessels, presence of stellate-shaped tumor cells, a less conspicuous inflammatory infiltrate, and an absence of the atypical hallmark cells characteristic of MIFS, even in high-grade examples with foci of pleomorphic sarcoma (conventional malignant fibrous histiocytoma). Cytogenetically, myxofibrosarcomas are associated with highly complex karyotypes lacking specific structural aberrations,^32,33 and, to date, no confirmed example has demonstrated t(1;10).

Our immunohistochemical analysis of tumor cells, which showed a rather nonspecific immunoprofile with vimentin, CD34, CD68, and focal actin expression, paralleled immunohistochemical data previously reported^{1,2,12,13,17,19,20} and indirectly supports electron microscopic findings^2,17 that the tumor cells of MIFS show features consistent with fibroblasts. Keratin ex-pression noted in 10 of our cases and in other MIFS studies^1,12,13,17 is not surprising as keratin has been de-tected in a variety of fibroblastic and myofibroblastic processes as well as in transformed fibroblasts in cul-ture.³⁴ We are the first to describe D2-40 im-munoexpression in MIFS. D2-40, an antibody directed against the 38-kDa transmembrane glycoprotein podo-planin,³⁵ is also a rather nonspecific immunomarker in the present context, as it is expressed in other myo-/fibroblastic tumors including benign fibrous histiocyto-ma, cellular neurothekeoma,^36,37 myxofibrosarcoma, and undifferentiated pleomorphic sarcoma (malignant fibrous histiocytoma).³⁸ We do not believe that MIFS is related to the D2-40-expressing tenosynovial giant cell tumor³⁹ or follicular dendritic cell tumor,⁴⁰ not only because of differing histologic features but also because, immunohistochemically, the key tumor cell of both tumors expresses clusterin^39,41 and, in tenosynovial giant cell tumor, desmin,⁴² both of which were virtually absent in our tumors. In a study of 18 cases of MIFS, Kovarik et al²⁰ claimed expression of CD163 and epidermal growth factor receptor in tumor cells in the majority of cases tested. However, in our experience, positivity for CD163, which is considered a specific macrophage marker,⁴³ is usually attributable to infiltrating reactive histiocytes and not to neoplastic fibroblasts. In contrast, epidermal growth factor receptor has been detected in soft tissue neoplasms of diverse histogenesis.^44,45

The overall recurrence rate of 51% in our study is higher than the mean recurrence rate of 30% calculated from large series published between 1998 and 2010 (in which 36 of 120 patients with follow-up data suffered recurrent disease).^{1–3,12,13,17–21} Our recurrence rate is closer to that reported by Meis-Kindblom and Kindblom² (67%) and probably reflects the preponderance of simple local excisions (without regard to margins) performed in both studies along with selection bias resulting from underdiagnosed tumors sent for consultation after recurrence. Although the lower recurrence rate described in some more recent studies is partly due to a short (<5y) follow-up interval,^18–21 improved treatment outcomes with recurrence rates ranging from 0% to 50% (mean, 13%),^1,13,18–23 as claimed in recent as well as some older studies, more likely reflect initial treatment by margin-negative, wide local, or radical surgical excisions.^1,18–21,23 Our results regarding the effect of surgery on recurrence rate agree with these latter findings. Moreover, 6 of the 12 patients in our study who experienced a recurrence but are presently disease free for over 5 years underwent wide local excision or amputation for the recurrent tumor.

As an adjunct to surgical treatment, radiation therapy had been attempted in only a small number of previously reported cases.^1,2 Recently, Tejwani et al²¹ reported an absence of recurrent disease in 13 patients treated with either preoperative or postoperative radiation, including 4 patients with positive surgical margins after initial excision, and concluded that radiation therapy assists in local control of MIFS. Although we did not statistically evaluate the clinical significance of this treatment modality, 4 of 6 patients receiving radiation after reexcision of a recurrence have remained free of disease for over 5 years, and 1 has been disease free for 12 months.

Four cases of MIFS with proven metastatic disease to regional lymph nodes (n = 3),^2,12,13 and lung,¹³ neck, and base of skull (1 each)¹³ have been documented in the literature. Three of these patients experienced local recurrence before metastatic spread.^2,13 Little histologic detail is provided for the majority of these cases, with the exception of the tumor reported by Sakaki et al¹² that metastasized to the lung and was described as hypercellular with a modestly high Ki-67 index of 12.1% and a mitotic rate of 6 mitoses per 50 HPF. Hassanein et al¹³ reported an MIFS metastatic to the lung with a Ki-67 index of 30. The one patient in our study who developed metastatic disease after suffering multiple recurrences had a primary tumor exhibiting both focal increased cellularity and high mitotic activity. On the basis of these limited data, we stress that conventional MIFS, particularly those that have recurred, and especially tumors with atypical histologic features as described above have a very low but definite risk for metastatic spread and should be managed accordingly.

Our study of MIFS reinforces the established view that MIFS is a low-grade inflammatory fibromyxoid neoplasm that predilects acral soft tissue and is composed mainly of modestly atypical plump spindled and epithelioid cells and a variably sized population of strikingly bizarre tumor cells. Our study also emphasizes the fact that MIFS is capable of exhibiting a wide breadth of morphologies ranging from tumors with focal hypercellularity, sarcoma-like vascularity, or increased or atypical mitotic activity simulating more aggressive sarcomas to lesions mimicking an inflammatory process because of an extreme paucity or frank absence of hallmark lesional cells. However, irrespective of histologic findings, a complete, preferably margin-negative, excision is the most effective approach for reducing the incidence of locally recurrent disease, which is the main cause of morbidity associated with the tumor. In addition, adjuvant radiation therapy may be considered for those patients with positive surgical margins, especially if the tumor harbors atypical histologic features.