Fetsch, John F. M.D.; Miettinen, Markku M.D.; Laskin, William B. M.D.; Michal, Michal M.D.; Enzinger, Franz M. M.D.
The purpose of this article is to present a series of 45 cases of an uncommon but distinctive pediatric soft tissue tumor, referred to here as a lipofibromatosis. This process appears to have a predilection for the hands and feet but also occurs in a wide variety of other sites. In the past, it was classified primarily as a variant of fibromatosis (especially congential and infantile/juvenile types) or fibrous hamartoma of infancy; however, consideration was also sometimes given to an early stage of calcifying aponeurotic fibroma and a variant of lipoblastoma. When these tumors are assembled as a group and compared with conventional examples of soft tissue tumors that enter into the differential diagnosis, it is apparent that separate classification is warranted. Data supporting this conclusion is presented.
MATERIALS AND METHODS
Archival material, accessioned to the Armed Forces Institute of Pathology (AFIP) between 1961 and 1992, was the sole source of cases for this study. Soft tissue tumors coded as a congenital, infantile, or juvenile fibromatosis (n [equals] 188), fibromatosis of the desmoid type, palmar-plantar type or not otherwise specified (n [equals] 503), fibrous hamartoma of infancy (n [equals] 46), lipoblastoma (n [equals] 55), juvenile fibrosarcoma (n [equals] 34), and mesenchymal hamartoma (n [equals] 1) were retrieved for evaluation. This study group includes every available case of fibromatosis (with the exception of some cases of infantile digital fibromatosis) from patients 20 years of age or younger, as well as a board sampling of fibromatoses throughout adulthood. Of the 827 tumors that were screened, 45 neoplasms with similar histologic features were identified and are the basis for this report.
All hematoxylin and eosin-stained sections were reviewed. In 15 cases, formalin-fixed, paraffin-embedded material was also examined with immunohistochemistry using the avidin-biotin complex (ABC) immunoperoxidase technique. Table 1 lists the immunohistochemical antibodies, their sources and dilutions. Enzymatic digestion at 37[deg]C for 3 minutes with Protease type VIII (Sigma Chemical Co, St. Louis, MO, USA; concentration [equals] .04 g/100 mL 0.1 M phosphate buffer, pH 7.8) was used to unmask antigenic sites before incubation with primary antibodies directed against keratins and epithelial membrane antigen. Tissue slides examined for CD34, Melan-A, and tyrosinase were microwaved in 10 mM sodium citrate buffer (pH 6.0) at or near boiling for 20 minutes and cooled for 45 minutes in the buffer before incubation with the primary antibodies. A similar microwave procedure using a 1mM EDTA solution (pH 8.0) was used on sections examined for BCL-2.
Primary antibodies were incubated with the histologic sections at room temperature for 1 hour. After incubation with a biotinylated secondary antibody, the immunohistochemical reaction was visualized using the avidin-biotin-peroxidase complex with either diaminobenzidine tetrahydrochloride or the Vector VIP substrate kit for peroxidase (Vector Laboratories, Burlingame, CA, USA) as the chromogen. Appropriate controls were tested simultaneously.
All of the sections with immunostaining were assigned a numerical value ranging from 0 to 4, depending on the percent of immunoreactive tumor cells. When staining was present in [le]10[percnt] of the tumor cells, 1 point was given. When more than 10[percnt] but [le]50[percnt] of the cells were immunoreactive, a value of 2 was assigned. Three points corresponds to [gt]50[percnt] but [le]75[percnt] of the cells being positive, and 4 points equals [gt]75[percnt] immunostaining.
Follow-up data was obtained by telephone conversation with the patients, their guardians, and physicians.
The clinical features of the study group are summarized in Table 2. There were 32 males, 12 females, and one person of unstated gender. The patients presented with a soft tissue mass involving the hand (n [equals] 18), arm (n [equals] 8), leg (n [equals] 7), foot (n [equals] 6), chest (n [equals] 3), abdomen (n [equals] 2), or head (n [equals] 1). Eight tumors were noted at birth. The lesions were generally described as painless and slow-growing. However, in one instance, rapid growth was reported.
The patients ranged in age from 11 days to 12 years (median, 1 yr) at the time of their initial surgical procedure. Information with regard to the type of initial management was available for 37 patients and included a biopsy (n [equals] 8) or attempt at local excision (n [equals] 29); the latter procedure was noted to be incomplete for five patients.
The differential diagnosis provided by the contributing pathologists included a lipomatous tumor variant (that is, a [ldquo]juvenile lipoma,[rdquo] fibrolipoma, lipoblastoma, spindle cell lipoma, or liposarcoma; n [equals] 21), a fibrous hamartoma of infancy (n [equals] 15), a fibromatosis variant (n [equals] 11), a mesenchymoma or mesenchymal hamartoma (n [equals] 6), a calcifying aponeurotic fibroma (n [equals] 4), a neuroma or neurofibroma variant (n [equals] 3), adipose tissue with reactive changes (n [equals] 2), and a lymphangioma and nodular fasciitis (n [equals] 1 each).
Gross and Microscopic Findings
A partial or complete gross description was available for 33 tumors. The lesions ranged in size from 1 to 7 cm in maximum dimension (median size, approximately 2 cm;Fig. 1). A yellowish or tan[ndash]white color and a rubbery, firm, or gritty consistency were usually evident. Fat or fibrofatty tissue was often specifically noted in the grossing record. Irregular, poorly defined margins were the rule; only five tumors were described as lobulated.
Histologic examination revealed abundant adipose tissue, typically comprising more than 50[percnt] of the specimen. Fat accounted for more than 70[percnt] of the resected tissue in 35 cases, whereas it accounted for [le]50[percnt] of the tissue in only three instances. Fascicles of fibroblastic cells traversed the adipose tissue, but in contrast with conventional fibromatoses, there was a striking association with fat septa (Figs. 2, 3, and 4). This helped to preserve the variably-sized fat lobules. The fibroblastic component generally contained a small or moderate amount of collagen, but in a few instances, more heavily collagenized foci were noted. Myxoid change was present in a few cases and was most pronounced in the specimen from the youngest patient, an 11-day-old infant (Fig. 5). In no instance did the lesions have [ldquo]wire-like[rdquo] ([ldquo]neurofibroma-like[rdquo]) fibrocollagenous connective tissue or the primitive myxoid mesenchymal nodules with concentric or whorled architecture as seen in fibrous hamartoma of infancy. Also, none of the tumors with sequential specimens showed evolution to a desmoid-type fibromatosis.
Most cases contained occasional small collections of univacuolated cells juxtaposed between the spindled fibroblastic component and the mature adipocytes (Fig. 6). These cells did not have true lipoblastic properties (that is, multiple lipid vacuoles surrounding and indenting the nucleus); some may have been degenerating adipocytes, whereas others could have been lipid-rich fibroblasts or a transitional stage between fibroblast and adipocyte.
The lesions were poorly marginated, and both the fat and fibrous elements infiltrated and entrapped regional structures, including vessels (n [equals] 45), nerves (n [equals] 44), skin adnexa (n [equals] 16), and skeletal muscle (n [equals] 18).
Mitotic figures were identified in 15 tumors. However, the maximum mitotic rate exceeded 5 mitotic figures per 50 high-power fields in only one instance. This case contained small areas with as many as 5 mitoses in 10 high-power fields, but the spindled element did not have excessive atypia or the cellularity or growth pattern of fibrosarcoma. In all cases, cytologic atypia was invariably mild, and there was little or no nuclear pleomorphism.
An additional subtle finding in two cases (patient nos. 2 and 3) was the presence of a small number of scattered pigmented, spindled, or dendritic cells, similar to that found in Bednar tumor (pigmented dermatofibrosarcoma protuberans), pigmented (melanotic) neurofibroma, and certain melanocytic nevi (Fig. 7). These cells were associated with the fibroblastic element in the subcutis and skeletal muscle, and were generally confined to the fat septa. Features that would warrant classification as a dermatofibrosarcoma protuberans, neurofibroma, or melanocytic nevus were absent.
Immunohistochemical data was collected from 15 cases. The spindle cell element of the tumors exhibited immunoreactivity for CD99 (10 of 12), CD34 (12 of 13), [agr]-smooth muscle actin (10 of 12), BCL-2 (7 of 12), S-100 protein (5 of 14), muscle actin (3 of 13), and EMA (2 of 10). Immunostaining for CD99 was typically moderate to diffuse with an average immunohistochemical score of 3 of 4. Immunostaining was less extensive for CD34 and BCL-2 with an average immunohistochemical score of 2. Immunoreactivity for [agr]-smooth muscle actin, muscle actin (HUC 1-1), S-100 protein, and EMA was generally limited with immunohistochemical scores ranging from 1 to 2. In no case did the fibroblastic element exhibit desmin (D33 and D-ER-11 clones) or keratin reactivity. Adipocytes in all of the tested specimens exhibited at least focal immunoreactivity for S-100 protein and BCL-2, and in 8 of 10 cases, they were focally reactivity for our cytokeratin cocktail. The rare pigmented cells that were present in two cases were positive for tyrosinase (2 of 2), Melan-A (2 of 2), HMB-45 (1 of 2), and S-100 protein (1 of 2).
Follow-up data, ranging from 2 months to 26 years (median and mean follow-up intervals, 6 yrs 7 mos and 8 yrs 7 mos, respectively), were available for 25 patients (56[percnt]). Eighteen patients (72[percnt]) had regrowth of their tumor or had persistent disease. Initial management of this group consisted of an attempted local excision (n [equals] 11), an unspecified procedure (n [equals] 3), and a biopsy (n [equals] 4). The excision was documented to be incomplete for three of the patients that had an attempted local resection. Of the four patients who initially had biopsy procedures, one underwent local excision of the area, followed by an amputation for recurrence. One had an additional biopsy, followed by a local procedure to alleviate discomfort, and then an apparent local excision, all occurring over a period of more than 6 years; the patient is currently lost to follow up. The third patient had an additional biopsy and multiple incomplete resections over a 20-year period; this individual had persistent but stable disease at last follow-up examination. The last patient had an additional biopsy, followed by two incomplete local excisions, and has been clinically free of disease for the past 5 years. Except for the single case noted above (patient no. 7), no one, to our knowledge, has been managed with radical surgery, and only one patient (patient no. 45) has received adjuvant radiation therapy.
There were seven patients with follow-up ranging from 9 to 26 years (median follow-up period, 19 yrs) who did not have regrowth of their tumor. Their management consisted of an attempted local excision (n [equals] 6) or an unspecified procedure (n [equals] 1). The attempt at local excision was reported to have been incomplete for two of the patients.
No tumor has given rise to metastases.
When patients who had regrowth of their tumor or persistent disease were compared with patients who had not experienced a recurrence, the former group was found to contain more congenital tumors, patients of the male sex, lesions involving the hands and feet, and documented incomplete procedures. From a histologic standpoint, the only discernible difference between the two groups was the more frequent presence of mitotic activity in the fibroblastic element of tumors that exhibited regrowth (53[percnt] vs 29[percnt]). Within the latter group, there were two tumors that exhibited 5 mitoses/50 high-power fields.
Tumor size, tumor depth, and the percent of intralesional fat were not found to influence behavior. However, it should be stressed that the sample size is relatively small.
Lipofibromatosis is an uncommon pediatric tumor that presents as a poorly demarcated mass involving the subcutis or deep soft tissues. Histologic examination reveals abundant adipose tissue traversed by bundles of spindled fibroblast-like cells. Although our findings demonstrate a wide anatomic distribution, a disproportionate number of cases involve the hands and feet.
While we would expect all observers to interpret the fibroblastic element as neoplastic, some might question whether fat should be regarded as an integral component of the process. We assert that it is based on the following observations. First, fat constituted a significant part of the clinical and gross mass, and its amount often appeared exaggerated when compared with what normally would be expected in the affected site. Second, the adipose tissue appeared more disorganized than normal with variably-sized, sometimes poorly demarcated, lobules. Third, the fat infiltrated, entrapped, and displaced normal tissue such as skeletal muscle. Finally, the distribution of the fibroblastic element within fat septa supports the notion that fat is integral to the process, because this pattern is a well-recognized feature of both benign and malignant lipogenic tumors that contain a nonlipogenic component.
Immunohistochemistry does not play a major role in the diagnosis of lipofibromatosis. The results we obtained were predictable with a few exceptions, the most important being the presence of focal (immunohistochemical score[colon] 1 or 2 of 4) S-100 protein in the spindled element of five of 14 tumors. The significance of this is unclear; in no instance was the pattern or extent of S-100 protein reactivity sufficient to warrant serious consideration of a nerve sheath or melanocytic tumor. While recruited S-100 protein-positive dendritic (antigen-presenting) cells were noted in a number of cases, these were discounted, and their presence was not sufficient to classify a tumor as positive for this marker. The presence of S-100 protein in some spindled tumor cells could reflect their potential to differentiate into fat, or might merely reflect a greater distribution for this calcium-binding protein than has generally been acknowledged. We have seen focal (although usually fairly limited) immunoreactivity for S-100 protein in the spindled element of a number of tumor types, including spindle cell lipoma, superficial angiomyxoma/cutaneous myxoma, 15 calcifying aponeurotic fibroma, 12 and synovial sarcoma, 14 to name a few.
Another unexpected finding warranting brief comment was the presence in two cases of a small number of pigmented (melanotic) cells with spindled or dendritic morphology, identical to the cells found in pigmented dermatofibrosarcoma protuberans, 7 pigmented variants of neurofibroma, 13 and some melanocytic nevi. The tumors in which these cells were found appeared identical in all other respects to the rest of the study group, and there was no compelling evidence to warrant classification in a different manner. In both instances, the tumors were congenital; one involved the forearm and the other involved the plantar surface of the foot. We suspect the pigmented cells are bystander cells that have become passively incorporated within the mass, because their normal migration pathway has been disturbed by the lipofibromatous process. 5,13
The differential diagnosis for lipofibromatosis includes juvenile (including congenital and infantile) fibromatosis, fibrous hamartoma of infancy, calcifying aponeurotic fibroma, and lipoblastoma.
Juvenile (infantile) fibromatosis is a designation that has been applied to a morphologically diverse group of locally aggressive soft tissue tumors composed of fibroblast-like cells. 2,9,10,21 Three subtypes have been recognized (the diffuse or mesenchymal, the fibroblastic, and the desmoid variants) with the belief that differences in histology are primarily attributable to the age of the patient and level of maturation. 10 However, it has not been conclusively established that all examples inevitably evolve into desmoid-like neoplasms. Therefore, while this classification scheme has proven useful from a management standpoint, it may not be pathogenetically legitimate. We mention this because during the course of our review, we encountered a number of tumors that were previously classified as a diffuse or fibroblastic variant of fibromatosis that either fulfilled our criteria or exhibited some overlapping features with the tumors classified here as lipofibromatosis. This said, it is important to point out that many examples of juvenile fibromatosis differ both clinically and morphologically from the tumors presented in this article. True fibromatoses typically exhibit more solid sheet-like growth and do not contain fat as an integral component. Although juvenile fibromatosis and lipofibromatosis have overlapping anatomic distributions, the former has a greater tendency to involve the fascia and musculature in the head and neck, shoulder girdle, upper arm, and thigh regions, whereas the latter seems to have a greater predilection for the hands and feet.
Fibrous hamartoma of infancy is a benign soft tissue lesion that usually presents in the first 2 years of life as a subcutaneous mass in the axilla, inguinal region, or proximal portion of an extremity. 6,8,16,19,20 This process typically contains four recognizable components[colon] (1) traversing bundles of spindled fibroblastic cells; (2) variable amounts of collagen, often with a hyalinized, [ldquo]wire-like[rdquo] appearance, similar to what is sometimes encountered in neurofibromas; (3) small nodular aggregates of immature-appearing stellate-shaped and spindled cells, often with a whirled arrangement and a myxoid matrix; and (4) mature univacuolated adipocytes. While some pathologists have diagnosed soft tissue lesions involving the hands and feet as fibrous hamartomas of infancy, others deny the existence of this process in these locations, or at least consider it a vanishingly rare occurrence. To get a better perspective on this controversy, we reviewed the AFIP soft tissue archives but found no fully typical example of fibrous hamartoma of infancy in either site. It is noteworthy, however, that some of the tumors included in our series as lipofibromatoses were interpreted at other institutions as fibrous hamartomas of infancy. Nevertheless, none of these lesions contained primitive organoid nests of mesenchymal cells set in a loose myxoid matrix or the peculiar neurofibroma-like fibrosis that is encountered in true fibrous hamartomas of infancy. It is conceivable, therefore, that if critically reappraised, some tumors of the hands and feet previously interpreted as fibrous hamartomas of infancy will reveal a morphology more in keeping with lipofibromatosis.
Calcifying aponeurotic fibroma is an uncommon mesenchymal tumor that shows a strong predilection for the hands and feet but occasionally involves other sites such as the lower back and knee region. 1,12,18 This process has a peak incidence in the first decade with almost half of all cases diagnosed by age 10. 12 The tumor has a recurrence rate more analogous to a fibromatosis than a fibroma, but it is morphology distinct from, and seems to be less clinically aggressive than, conventional fibromatoses. When biopsied in its early stages, the characteristic nodular foci with calcification and cartilage formation may be absent, and it can be difficult to distinguish from a juvenile fibromatosis or lipofibromatosis. However, the presence of small epithelioid fibroblasts in a corded array and the lack of a defined fatty element are helpful distinguishing features.
Lipoblastoma(tosis) is another soft tissue tumor of infancy and early childhood. 3,4,11 It has a wide anatomic distribution but chiefly affects the extremities. Conventional examples consist of variably-sized fat lobules, with or without myxoid change and cellular immaturity, that are separated by septa of varying thickness. Most examples have low septal cellularity and are easily distinguished from lipofibromatosis. However, rare purported examples of this tumor have been described with broad moderately cellular septa, and these can be difficult to distinguish from a lipofibromatosis or juvenile fibromatosis. 11 The clinical circumstances, amount of fat, and the appearance of the fibroblastic element (for example, the extent of the fibroblastic proliferation, its cellularity, and whether fascicular growth is noted) may help to resolve this differential diagnosis.
In the course of our literature review, we encountered one publication using the designation [ldquo]lipofibromatosis.[rdquo]17 However, this was in the context of a congenital abnormality associated with macrodactyly that has also been referred to as macrodactylia fibrolipomatosis, macrodystrophia lipomatosa, megalodactyly, dactylomegaly, and local gigantism. Histologically, the authors describe abundant fibrofatty tissue that [ldquo]was normal in all aspects,[rdquo] except for an increase in amount. Thus, this appears to be a different process from what we have described and might be better designated by an alternate term.
In summary, we have described a group of 45 pediatric tumors with a wide anatomic distribution but an apparent predilection for the hands and feet. Classification of these tumors has in the past been controversial with most examples being considered a form of juvenile fibromatosis or variant of fibrous hamartoma of infancy. We view the clinicopathologic features as distinctive and interpret fat as an integral element of this process. The designation [ldquo]lipofibromatosis[rdquo] is preferred over [ldquo]fibrolipomatosis[rdquo] (even though fat often dominates the process) because the key proliferative element appears to be the fibroblastic component. Tumors of this type have a predilection for local recurrence when incompletely excised. As a result, complete removal is preferable, especially if it can be accomplished with minimal morbidity. However, in cases in which significant functional compromise will be the end result of complete excision, management must be individualized because there are some cases with long-term follow-up that have not progressed even though the process was incompletely excised. This may be the result of lesional maturation, analogous to that described in other benign fibrous and lipomatous tumors.
[copy] 2000 Lippincott Williams [amp] Wilkins, Inc.