The degree of sclerosis is associated with prognosis in well-differentiated liposarcoma of the retroperitoneum

Abstract

Background and Objectives:

Well-differentiated liposarcomas (WDL) are often partly composed of sclerotic tissue, however, the amount varies widely between tumors, and its prognostic significance is unknown. We hypothesized that tumors with more sclerosis would behave more aggressively.

Methods:

Primary retroperitoneal WDL from 29 patients resected at our institution with follow-up were histologically evaluated by soft tissue pathologists blinded to outcome. Tumors with ≥ 10% sclerosis were designated “sclerotic” while tumors with < 10% sclerosis were designated as “minimally sclerotic”. Cellular and dedifferentiated tumors were excluded. Clinical parameters and radiologic assessments on computed tomography (CT) were recorded.

Results:

Histological evaluation identified 13 minimally sclerotic WDL and 16 sclerotic WDL. Median follow-up was 9 years (range, 3–20). Median recurrence-free survival (RFS) and median overall survival (OS) were 6.16 and 13.9 years, respectively. Compared with patients with sclerotic WDL, those with minimally sclerotic WDL had superior RFS (HR = 0.17 [95% CI, 0.06–0.53], P = .002) and OS (log-rank test, P = .002). Sclerotic WDL exhibited higher Houndsfield Units than minimally sclerotic WDL (26 vs 1, P = .040).

Conclusions:

Minimally sclerotic WDL were associated with more favorable outcome compared with sclerotic tumors. Assessment of sclerosis in WDL is likely a useful prognostic marker.

1 |. INTRODUCTION

Atypical lipomatous tumor/well-differentiated liposarcomas (ALT/ WDL) are low-grade malignant lipomatous neoplasms that frequently involve the retroperitoneum and often follow a protracted course with multiple recurrences. These tumors are classically characterized by mature adipocytes, fibrous bands, atypical stromal cells, and the presence of recurrent double minute, supernumerary ring, and giant chromosomes associated with punctuated amplification across the 12q13–15 region that includes MDM2, CDK4, and HMGA2.¹

ALT/WDL have a broad range of histological appearances leading to the description of multiple subtypes.² Enterline et al originally described the lipoma-like variant stating, “[t]he major portion of tumors of this type can be easily mistaken for lipoma or for normal fat…”³ In contrast, the sclerosing type is characterized by broad sheets and bands of collagen, the composition of which ranges from finely fibrillary to densely sclerotic, dividing variable amounts of adipose.⁴ Scattered pleomorphic stromal cells and floret-like cells are typically prominent, particularly within sclerotic zones. The inflammatory variant is composed of adipose, hypocellular fibrous stroma with scattered atypical cells, and a dense neutrophilic, lymphoplasmacytic, or mixed infiltrate that can obscure the essential nature of the tumor. This can lead to a mistaken impression of lymphoma (Hodgkin and non-Hodgkin types) or other lymphocyte predominate lesions such as Castleman disease or inflammatory myofibroblastic tumor.^5–7 ALT/WDL can also exhibit prominent myxoid stroma and, when combined with short uniform spindle cells and a plexiform vascular network, these tumors can mimic myxoid liposarcoma.^8–10 However, most tumors show a mixture of adipocytic, fibrous, inflammatory, and myxoid components, and classic examples of ALT/WDL subtypes represent extremes on a multiaxis spectrum. Other features that can be seen in ALT/WDL include bone formation,¹¹ hibernoma-like areas,^8,12 and smooth muscle differentiation.^13,14

While the more aggressive nature of dedifferentiation is well-established, it is generally believed that the extent of sclerosis, myxoid stroma, and inflammation in ALT/WDL does not influence prognosis, although there are limited data to support this assertion. Recently, a retrospective review of pathology reports at our institution showed that ALT/WDL with qualifiers (noted to have morphological features such as sclerosis, myxoid change, etc) were associated with worse clinical outcomes and appeared to behave more aggressively.¹⁵ This prompted us to undertake a standardized and systematic histologic and radiographic evaluation of the most common histological feature, sclerosis, in a series of primary retroperitoneal ALT/WDL to understand its prognostic significance.

2 |. MATERIALS AND METHODS

Institutional review board approval was obtained, and the surgical pathology database was examined for primary retroperitoneal/abdominal ALT/WDL (1996–2011) with available hematoxylin and eosin (H&E) stained slides and at least 34 months follow-up. Slides were independently reviewed by four pathologists (WLW, AJL, JSAC, and TA) with experience in soft tissue pathology, as well as in consensus. All cases showed distinctive diagnostic histologic features of ALT/WDL often with additional demonstration of 12q15 amplification by fluorescence in-situ hybridization. Sclerosis was strictly defined as hypocellular fibrous nonlipogenic areas. A cut-off of 10% sclerosis was selected to separate cases that were minimally sclerotic (lipoma-like) vs sclerotic. After examination of all available H&E slides (typically at least one section per cm of the greatest dimension of the tumor), each case was categorized as “minimally sclerotic (lipoma-like)” (composed of mature-appearing adipocytes with < 10% sclerosis, Figure 1) or “sclerotic” (≥ 10% sclerosis, Figure 2). ALT/WDL displaying extensive inflammation (≥ 10% as described above) and/or extensive myxoid stroma (≥ 10% as described above) were also explored. However, these two groups contained too few cases for meaningful multivariate analysis and thus were not included in this study. Cases with hypercellular areas, termed cellular ALT, (also referred to by some as low-grade dedifferentiation) or dedifferentiated foci, as defined by Evans,⁸ were also excluded. Clinical features were noted, including: age at diagnosis, gender, tumor focality, surgeon’s clinical assessment of the completeness of resection (R0/R1 vs R2), nephrectomy, nonrenal organ resection, all-cause mortality, and disease recurrence. Imaging assessment was performed by a subspecialty-trained radiologist (BA) blinded to pathology and clinical outcomes and consisted of (a) subjective assessment of the overall lipoma-like appearance of the lesion, (b) subjective assessment of percentage of the lesion composed of higher density (nonfat) components (≥ 10% vs < 10%), and (c) the average attenuation (measured in Hounsfield units [HU]) of the densest, noncalcified portion of the lesion. All assessment and measurements were performed on Philips IntelliSpace PACS (Philips Healthcare, Best, Netherlands). Computed tomography (CT) imaging was performed on a variety of devices both inside and outside our institution.

FIGURE 1.

Representative images of minimally sclerotic (lipoma-like) well-differentiated liposarcoma. A and B, H&E slides show mostly clear adipocytic tumor cells. C, Higher power examination (x100) reveals variable-sized adipocytes with scattered atypical hyperchromatic cells (inset, x400) well as around vessels, D, (x200). H&E, hemotoxylin and eosin

FIGURE 2.

Representative images of sclerotic well-differentiated liposarcoma. A and B, H&E slides show evident eosinophilic fibrosis. C and D, Higher power examination reveals collagenous bands separating mature adipocytes with scattered atypical hyperchromatic stromal cells (x200 and x400). H&E, hemotoxylin and eosin

The Wilcoxon rank sum test¹⁶ was used to compare continuous variables between histologic subtypes. The χ² test and Fisher’s exact test¹⁶ were used to determine whether proportions of categorical variables were equal between histologic subtypes. Analyses of recurrence-free (RFS) and overall survival (OS) were performed. RFS was defined as from the time of diagnosis to the time of recurrence or death, whichever occurred first or to the time of last contact. OS was defined as from the time of diagnosis to the time of death or to the time of last contact. The distributions of RFS and OS were estimated by the Kaplan-Meier method.¹⁷ Log-rank test¹⁸ was performed to test the difference in survival between patient characteristics groups. Regression analyses of survival data based on the Cox proportional hazard (PH) regression model¹⁹ were conducted on RFS and OS. Functional form was assessed and proportional hazard assumption was tested. Multivariate model with backward elimination method was fitted. Firth correction²⁰ was applied as necessary. A two-sided P-value < .05 was considered statistically significant. All statistical analyses were performed in SAS 9.4 (SAS Institute Inc., Cary, NC).

3 |. RESULTS

Twenty-nine patients with primary retroperitoneal ALT/WDL resected at our institution with at least 34 months of clinical follow-up and histological slides available for review were identified (Table 1). There were 12 women and 17 men with a median age at diagnosis of 56 years (range, 22–80) and a median follow-up time of 9 years (range, 3–20). Thirteen patients presented with minimally sclerotic (lipoma-like) ALT/WDL and 16 presented with sclerotic tumors per our defined histologic criteria. Complete surgical excision was planned and attempted for all patients. Multifocality was not observed. No significant difference was seen between patients with minimally sclerotic (lipoma-like) ALT/WDL and those with sclerotic ALT/WDL with respect to age at diagnosis (61 vs 54 years, P = .614), sex (38 vs 44% female, P = .774), tumor size (25 vs 31 cm, P = .204), subsequent development of dedifferentiation (8 vs 6%, P = 1.00), and nephrectomy (23 vs 31%, P = .697). The median follow-up time was longer in minimally sclerotic (lipoma-like) group (10 years) compared with the sclerotic group (8 years). A lower proportion of nonrenal organ resections (8% vs 31%, P = .183) and R2 resections (0%vs 19%, P = .232) was noted in the minimally sclerotic (lipoma-like) group compared with the sclerotic group, but these differences were not statistically significant.

TABLE 1.

Summary of clinical, pathological, and radiological findings

n	Total 29	Lipoma-like 13 (45%)	Sclerotic 16 (55%)	P-value
Gender (M:F)	12:17	8:5	9:7	.774
Median age (range) (y)	56 (22–80)	61 (22–76)	54 (38–80)	.614
Median tumor size (range) (cm)	30 (9–100)	25 (9–100)	31 (11–78)	.204
Resection status (RO/R1: R2)	26:3 (10%)	13:0 (0%)	13:3 (19%)	.232
Nephrectomy	8 (28%)	3 (23%)	5 (31%)	.697
Nonrenal organs removed	6 (21%)	1 (8%)	5 (31%)	.183
HU of densest, noncalcified (range)	21 (−55–55)	1 (−55–38)	26 (8–55)	.040
Radiological impression of lesion ≤10% dense	5/20 (25%)	3/8 (38%)	2/12 (17%)	.347
Overall radiological impression by CT as lipoma-like	6/20 (30%)	4/8 (50%)	2/12 (17%)	.161
Median follow-up (range) (y)	9 (3–20)	10 (5–20)	8 (3–14)
Developed differentiation	2 (7%)	1 (8%)	1 (6%)
Local recurrence	17 (59%)	4 (31%)	13 (81%)
Median time to first recurrence (range) (y)	4 (1–10)	6 (2–10)	3 (0–10)
Dead at last follow-up	9 (31%)	2 (15%)	7 (44%)
Disease at last follow-up	16 (55%)	3 (25%)	13 (81%)

Abbreviations: CT, computed tomography; F, female; HU, Hounsfield units; M, male.

Minimally sclerotic (lipoma-like) ALT/WDL were associated with better patient survival outcomes. Four of 13 (31%) patients with minimally sclerotic (lipoma-like) ALT/WDL recurred, while 13 of 16 (81%) patients with sclerotic tumors recurred. Median RFS in this study was 6.2 years. However, median RFS for patients with minimally sclerotic (lipoma-like) ALT/WDL was 13.9 (95% CI, 6.1, not reached) years vs 3.5 (95% CI, 1.9, 6.3) years for those with sclerotic ALT/WDL (log-rank test, P < .001, Figure 3). Furthermore, in a multivariate Cox proportional hazard regression model, which included tumor size, minimally sclerotic (lipoma-like) histology was significantly associated with longer RFS (HR = 0.14 [95% CI, 0.04, 0.47], P = .001).

FIGURE 3.

Patients with primary minimally sclerotic “lipoma-like” retroperitoneal liposarcoma have longer, A, recurrence-free and B, overall survival compared with patients with sclerotic tumors

Minimally sclerotic (lipoma-like) ALT/WDL was also associated with better overall survival. Two of 13 (15%) patients with minimally sclerotic tumors died, compared with 7 of 16 (44%) patients with sclerotic ALT/WDL. Median OS in this study was 13.9 years. OS for patients with minimally sclerotic (lipoma-like) ALT/WDL was longer compared with those with sclerotic ALT/WDL (not reached [95% CI, 13.5, not reached] vs 10.7 [95% CI, 5.7, not reached] years, log-rank test, P = .002; Figure 3). The difference in OS between minimally sclerotic (lipomas-like) and sclerotic groups in the univariate Cox PH model was not significant (P = .05) in this small study. Given that the number of death events was small, multivariate Cox PH regression analysis was not performed for OS.

Resection status and tumor size were also associated with outcome. Patients with R2 resections had inferior RFS and OS compared with patients with R0/R1 resections (log-rank tests, P = .002 and P < .001, respectively). Each centimeter increase in tumor size increased the risk of recurrence by 3% (HR: 1.03 [95% CI, 1.00, 1.05], P = .016) after controlling for sclerosis. However, tumor size was not significantly associated with OS (HR = 1.02 [95%, 1.00, 1.04], P = .059). Age at diagnosis, sex, nephrectomy, and nonrenal organ removal were not associated with RFS or OS in univariate Cox regression.

Twenty cases had available preresection CT images (Figure 4). Five cases were determined to have ≥ 10% nonfat component dense areas by CT imaging, and six cases were determined to have < 10% nonfat dense areas by overall radiological impression. Subjective classification of radiologic subtype correlated with the histologic sclerosis group classification in some cases but not all. There was a trend toward correlation with histological assessment and overall assessment of lipoma-like appearance (P = .161), less so by the classification with nonfat component using a 10% cut-off (P = .347) with radiological studies, though neither reached statistical significance. Sclerotic tumors were found to be denser compared to minimally sclerotic (lipoma-like) tumors (HU of the densest, noncalcified part of the tumor: 26 vs 1 HU, P = .040). Radiologic estimations of subtype and HU of the densest portion of the lesion were not associated with RFS or OS in univariate Cox regression.

FIGURE 4.

Representative computed tomography images in this series. A, Lipoma-like well-differentiated liposarcoma composed of predominantly fat and only few fibrous bands. The highest attenuation region in this lesion had mean attenuation of − 12.5 HU. B, Sclerotic well-differentiated liposarcoma with a large, high-attenuation component. The highest attenuation region in this lesion had mean attenuation of 35 HU. Sclerotic well-differentiated liposarcomas tended to have higher HU in the densest, noncalcified portion of the tumor than lipoma-like tumors. HU, Hounsfield units

4 |. DISCUSSION

In our series, patients with minimally sclerotic (lipoma-like) ALT/ WDL had a better prognosis than patients with sclerotic tumors when accounting for other known prognostic factors. Specifically, patient age, tumor site, tumor focality, extent of resection, and margin status, which have been included in a recently developed and validated nomogram,^21,22 did not vary significantly between the minimally sclerotic (lipoma-like) and sclerotic groups in our study. Recent data also suggest heterologous myogenic differentiation has prognostic significance in ALT/WDL and dedifferentiated liposarcomas.²³ However, myogenic differentiation was not noted in any of our cases.

While limited, data in the past have hinted at the prognostic significance of sclerosis in ALT/WDL. Evans et al noted that the absence of prominent fibrous areas seemed to be a favorable prognostic factor among patients with ALT/WDL,²⁴ and in Enterline et al original series, the OS of the lone patient with a retroperitoneal lipoma-like ALT/WDL was 29 years.³ In contrast, Singer et al examined sclerosing versus lipoma-like ALT/WDL and found no difference in three-year disease-specific survival; however, a very different cut-off of ≥ 25% sclerosing was used.²⁵ We analyzed RFS and OS with longer follow-up and set a lower cut-off of ≥ 10% sclerotic areas, and minimal sclerosis was found to be an independent favorable prognostic factor. In practice, the majority of cases showed negligible sclerosis. This suggests that a strict cut-off for sclerosis should be used and that even relatively small proportions of sclerosis portend a more locally aggressive course.

It is not entirely clear why tumors with increased sclerosis would be more aggressive. One explanation is that the degree of sclerosis in ALT/WDL reflects clinically important differences in underlying tumor biology. Histologically, fibrous/sclerotic areas in ALT/WDL appear less differentiated (not adipocytic) or perhaps transdifferentiated (fibrous or “stromal” differentiation). In many tumor types, the degree to which a tumor recapitulates normal tissue reflects the underlying biology and thus correlates with behavior. This concept of tumor differentiation has been incorporated into modern sarcoma grading.²⁶ Thus, one might expect minimally sclerotic (mostly adipocytic) tumors to exhibit less aggressive behavior compared with sclerotic tumors. However, the rate of dedifferentiation in subsequent recurrences between the two groups did not significantly vary, though the small number of events (n = 2, 1 in each group) makes reliable analysis difficult. Another possibility is that sclerotic tumors might be more adherent and infiltrative to adjacent structures resulting in a more challenging resection. While a higher proportion of patients with sclerotic tumors underwent nephrectomy (23% vs 31%) and nonrenal organ resection (8% vs 31%), and had higher rates of R2 resections (0% vs 19%), these differences were not statistically significant. Thus, the basis for the more favorable prognosis observed in minimally sclerotic (lipoma-like) ALT/WDL remains speculative.

While overall assessment on imaging did not reliably classify lesions to the correct histologic group, quantitative assessment of the attenuation (HU) of the densest aspect of the lesion was significantly associated with the histologic group. However, given there was some overlap in the HU of the two groups, further work is needed to develop reliable imaging correlates.

This study is limited by small sample size and by its retrospective design. Only primary tumors resected at our institution were included, which significantly limited acquisition as patients are often referred to our center with recurrences. Further, it would have been preferable for all specimens to be grossly evaluated prospectively to ensure the histologic sections were fully proportionately representative of the overall specimen. Fortunately, in routine practice, more solid areas are generally sampled preferentially to evaluate for dedifferentiation. Therefore, grossly solid appearing areas (ie, sclerotic, myxoid, and inflammatory components) tend to be overrepresented in histologic sections. We therefore suspect tumors with < 10% sclerotic areas (or other nonadipocytic components) microscopically are truly minimally sclerotic and as previously mentioned, the majority of tumors defined as minimally sclerotic were easily recognized as lipoma-like with only scattered delicate fibrous bands and few atypical stromal cells.

5 |. CONCLUSION

In summary, minimally sclerotic (lipoma-like) retroperitoneal ALT/WDL were associated with longer OS and RFS compared with sclerotic retroperitoneal ALT/WDL. Larger studies are needed to independently verify these findings, including testing for improved prediction within validated nomograms, and further inquiry is needed to determine the prognostic significance of the other histologic components (ie, inflammatory and myxoid) and tumors with distinctive features.

Abbreviations:

ALT

Atypical lipomatous tumor

CT

Computed tomography

HU

Hounsfield units

OS

Overall survival

PH

Cox proportional hazard

RFS

Recurrence-free

WDL

Well-differentiated liposarcoma