Myeloid Sarcoma Expressing Keratins and Mimicking Carcinoma—Case Report and Literature Review

Abstract

Unusual presentations of otherwise common hematopoietic neoplasms are a well-recognized diagnostic challenge. Herein, we present a case study of a previously healthy 64 year old woman with myeloid sarcoma whose diagnosis was delayed by an unusual immunohistochemical staining pattern, including cytokeratin expression, by the neoplastic cells and by possible anchoring bias introduced by radiographic and flow cytometric immunophenotyping reports. This case study emphasizes the need to integrate clinical, radiographic, histologic, and immunophenotyping data for rapid and accurate tissue diagnoses while being wary of the lack of specificity for many common immunophenotypic markers.

Unusual presentations of otherwise common hematopoietic neoplasms are a well-recognized diagnostic challenge.¹ The challenge is compounded when there is overlap between the immunohistochemical staining patterns of hematopoietic and nonhematopoietic neoplasms. Anchoring bias, the tendency to rely too heavily on an initial piece of information, and overreliance on radiographic or other laboratory findings to guide the microscopic pathology workup can introduce diagnostic errors and delays.² Herein, we present a patient whose ultimate diagnosis of myeloid sarcoma was delayed by both an unusual pattern of immunohistochemical staining and bias introduced by suspicion for lymphoma described in radiologic and flow cytometric immunophenotyping reports.

Clinical History

A previously healthy 64 year old female patient presented to her primary care physician with a painful right inguinal lump associated with swelling in her right leg that had appeared over the previous week. Review of systems was essentially negative. On physical exam, the woman was described as appearing well, with normal vital signs including normal oxygen saturation levels. Notable findings included a palpable inguinal mass and leg swelling. Laboratory studies including an automated complete blood count with leukocyte differential, routine chemistries, liver function tests, and urinalysis were all within normal limits. A computed tomography scan of the abdomen and pelvis performed with contrast showed right-sided inguinal (up to 10 cm), pelvic, retroperitoneal, and axillary lymphadenopathy. The radiologist interpreted the findings as concerning for lymphoma.

Right inguinal lymph node biopsy was performed immediately after the radiographic studies. The local pathologist examined a frozen tissue section, requested additional tissue, and forwarded fresh tissue to a reference laboratory with instructions that paraffin blocks were to be sent to the same reference laboratory the following day. Fresh tissue analyzed by flow cytometry at the reference laboratory identified a population of cells that was suspicious for, but not diagnostic of, a clonal B-cell population. The report noted the possibility of B-cell lymphoma, with the caveat that interpretation was limited by poor cell viability.

A final tissue diagnosis was delayed for some weeks. Although the patient was initially managed symptomatically as an outpatient, development of symptomatic anemia, thrombocytopenia, coagulopathy, and elevated serum lactate dehydrogenase (LDH) prompted transfer to our institution.

At the patient’s admission to our hospital, we did not have access to the original diagnostic material but were able to see the outside pathology report, which described a metastatic high-grade malignant neoplasm expressing multiple cytokeratins and S-100 by immunohistochemistry (IHC) and offered a differential that included melanoma and poorly differentiated carcinoma. In addition, our own clinical team noted the possibility of B-cell lymphoma, primarily based on a review of outside electronic medical records.

Because the original tissue biopsy was not readily available for our review, needle core biopsy of an enlarged right inguinal lymph node was repeated, with specimens sent for permanent sections and flow cytometry. On review of our flow cytometry and IHC, we verbally communicated our findings to the clinical team and recommended evaluation of the patient’s bone marrow. On the seventh hospital day, bone marrow biopsy confirmed the diagnosis of myeloid sarcoma with bone marrow involvement by myeloid leukemia. However, by this time, the patient had developed severe lactic acidosis. At the request of her family, life support was removed and the patient died.

Laboratory Evaluation

Lymph Node Needle Core Biopsy

Hematoxylin and eosin (H&E) stained sections of the lymph node core biopsy showed complete effacement by large atypical cells with morphologic characteristics suggestive of blasts, and focal necrosis (Image 1, A). Immunohistochemical stains were positive for the cytokeratin markers AE1/AE3 and CK8/18; positive for BCL-6, CD43, CD4, and lysozyme; and equivocal for S-100 (very weak staining of cytoplasm only with this antibody). Staining was negative for CD45 and CD30 (Image 1, B–J) and for CD3, other T-cell markers (CD2, CD5, CD7, and CD8), CD20, CD34, CD68 KP-1, CD117, CD123, CD138, PAX5, and MPO (not illustrated). Flow cytometric immunophenotyping performed on tissue from the needle core biopsy using B- and T-cell tubes was noncontributory because of poor cell preservation.

Peripheral Blood and Bone Marrow Aspirate

Rare neoplastic cells were identified on a blood smear (Image 2). However, flow cytometric immunophenotyping performed on blood using B-cell, T-cell, and myeloid tubes was again noncontributory, likely a result of the small number of circulating neoplastic cells present.

A Wright-Giemsa stained bone marrow aspirate smear showed trilineage hematopoiesis with dysgranulopoiesis and a high proportion of atypical blasts (Image 3). The bone marrow had abundant viable neoplastic cells, so a sufficient number of cells survived processing to form a discrete population for analysis by flow cytometric immunophenotyping. Overall, flow cytometry showed an abnormal cell population with high forward-scatter expressing bright CD4 and CD45 but lacking CD56 (Figure 1). However, the size/complexity of the cells combined with some degree of autofluorescence made it difficult to reliably interpret antigen expression/intensity for all antigens tested. Therefore, for the purposes of this report, only select antigens from the T-cell tube are presented because we could confidently distinguish the expression of these antigens from nonspecific staining of nonviable events.

The H&E and immunoperoxidase stains for the bone marrow clot section showed focal and interstitial marrow infiltration by atypical blasts. In addition to immunohistochemical stains performed on the lymph node needle core biopsy, CD33 immunoperoxidase stain was performed on the bone marrow clot section (Allina Laboratories, Minneapolis, MN). Blasts in bone marrow clot sections showed the expression of CD43, lysozyme, and CD33 (Image 4). Neoplastic cells in bone marrow trephine biopsy sections also showed the expression of BCL-6, CD4, and weak CD45 (not illustrated). The discrepancy between bright CD45 expression as detected by flow cytometry and weak expression as interpreted by IHC was attributed to the use of different monoclonal antibodies (RPD/18 for IHC and HI30 for flow cytometry). Slides for this patient’s original lymph node biopsy were not available for review.

Discussion

Herein, we describe a patient with acute myeloid leukemia (AML) presenting as myeloid sarcoma in a lymph node where the ultimate diagnosis was delayed and the outcome was poor. The consulting pathologists’ comments in the initial lymph node biopsy report raise the possibility that diagnostic delays and interpretations of possible carcinoma, melanoma, and lymphoma may have resulted from a combination of the unusual immunoperoxidase staining pattern with keratin and S100 expression by the tumor cells as interpreted by their laboratory, along with anchoring bias based on the reported flow cytometric immunophenotyping.

This case study highlights common diagnostic challenges for pathologists that all practicing physicians need to be aware of. Unusual presentations of otherwise well-recognized and treatable malignancies can lead to delayed and/or inaccurate diagnoses, particularly when:

• -

  There is an unusual pattern of immunoperoxidase staining, as was seen in this patient.

• -

  There is anchoring bias based on suspicions described in radiology or other laboratory reports.

Myeloid sarcoma (de novo presentation of AML outside of the bone marrow) presenting in a lymph node is a well-recognized diagnostic challenge.¹ Expedient and accurate diagnosis is critical; large series and case reports have established that myeloid sarcoma can be highly treatable and even curable.^3,4 Rapid and accurate diagnosis of any malignancy in the current era requires clinicopathologic correlation and the ability to discern useful from nonuseful data.

A careful review of the medical literature reveals a substantial overlap in immunostaining patterns between hematopoietic and nonhematopoietic neoplasms (Image 5). Nonspecific/aberrant staining of myeloid neoplasms in addition to T-cell lymphomas with cytokeratin markers in paraffin tissue sections has been described in the medical literature.^5-8 However, there are only a handful of case reports describing myeloid sarcomas that express cytokeratins.^9,10 In addition, recently published algorithms for the use of immunoperoxidase stains to accurately diagnose poorly differentiated large-cell neoplasms may not fully address CD45-negative hematopoietic neoplasms that express cytokeratins.¹¹

In this case study, we were guided to a diagnosis of myeloid sarcoma/AML upon detecting a strong surface expression of both CD4 and CD43 in the absence of other T-lineage markers. Uniform surface staining of neoplastic cells by CD43 has been shown to be specific and sensitive for the detection of hematopoietic neoplasms. However, focal weak nuclear staining has been described for some carcinomas,¹² and CD43 has not been described as being expressed in all reported patients with myeloid sarcoma.¹³ The CD4 expression appears specific for neoplasms of hematolymphoid origin and has not been described as being expressed for any cell types other than those of T-cell, histiocytic, or myeloid lineage. Histiocytic sarcoma should also be considered in the differential diagnosis of undifferentiated large-cell neoplasms that express S-100 protein along with lysozyme and cytoplasmic CD4. Our patient’s condition can be distinguished from having histiocytic sarcoma by the expression of the myeloid marker CD33 and the absence of staining for CD68, as described in the current World Health Organization classification.¹⁴

Flow cytometric immunophenotyping represents a powerful diagnostic tool that has become standard for evaluating specimens from lymph nodes, blood, bone marrow, and/or any tissue specimen suspected of harboring a hematopoietic neoplasm. Nonetheless, published series of flow cytometric analyses presented by respected reference laboratories have identified patients suspected of having a diagnosis or classification of a lymphoid neoplasm that was subsequently excluded by microscopy and IHC.^15,16 Notably, pathologists or other practitioners not specifically trained or experienced in the interpretation of flow cytometric immunophenotyping may be unaware of the limitations of this modality.

Anchoring bias, defined as an overreliance on the initial information offered, may have played a role in the delay of a specific diagnosis for the patient described in this report.² Although the authors did not have the opportunity to review this patient’s original excisional lymph node biopsy, both clinical physicians and pathologists at our institution were expecting to confirm the outside diagnosis based on the original radiographic, flow cytometry, and surgical pathology reports that were available in the electronic record. It was only when flow cytometric immunophenotyping performed on a repeat lymph node needle core biopsy showed low viability and autofluorescence that we paused to consider unusual presentations of neoplasms in lymph nodes and gathered additional information before making a specific diagnosis.

Conclusion

A previously healthy woman presented to her primary care practitioner with concerning findings on physical exam. A rapid clinical laboratory and radiographic workup with diagnostic lymph node biopsy were performed by her community health system. Unfortunately, deadly complications occurred while the diagnosis of her potentially treatable neoplasm was delayed, with an unusual immunohistochemical staining pattern and anchoring bias as likely contributors. We emphasize the importance of the integration of clinical, radiographic, histologic, and immunophenotypic data to arrive at a comprehensive diagnosis for the patient.

Abbreviations

Abbreviations
 
• IHC

  immunohistochemistry

 
• H&E

  hematoxylin and eosin

 
• AML

  acute myeloid leukemia

References