記住我
Abstract
Background: Effusions as part of hematologic neoplasms are rare and as a primary presentation, rarer. In standalone laboratories of developing countries, resorting to techniques such as flow cytometry or immunohisto/cytochemistry may not be possible. A near definitive diagnosis on cytomorphology would, therefore, be an ideal beginning. To that end, we compiled our cases of primary hematolymphoid effusions, devising reproducible reporting categories and looked at their concordance with the final histopathology. Subjects and Methods: Fifty-four cases of primary hematolymphoid effusions over 10 years with cytology-histopathology correlation were chosen. Post morphology assessment, the cases were organized into six categories: suspicious of hematolymphoid malignancy, non-Hodgkin lymphoma-high-grade (NHL-HG), low-grade NHL (NHL-LG), Burkitt lymphoma, acute leukemias, and plasma cell dyscrasias. Discordance with histology was assigned as major and minor based mainly on therapeutic implications. Results: Concordance was seen in a good number (81.5%) of cases. The NHL-HG and NHL-LG categories contributed to 33.3% each of major discordance. Tuberculosis and epithelial malignancies comprised the bulk of the major discordance. Overdiagnosis of a high-grade lymphoma for a histologically proven low-grade follicular lymphoma was the only case with minor discordance. Conclusion: The cytologic categories used are not foolproof for hematologic neoplasms but have a fairly good concordance. A scanty abnormal population should always be viewed with suspicion and definitive labels should be avoided. While morphologic examination is fraught with danger, a good assessment directs the judicious selection of ancillary methods, and hence cannot be supplanted.
Keywords: Cytology, cytomorphology, effusion, hematologic neoplasms, NHL
How to cite this article: Introduction 
Hematolymphoid malignancies (HLM) comprise only a miniscule number (<1%) of all effusions received in a cytopathology laboratory.[1],[2],[3] Overall, it is acknowledged that body fluid involvement is uncommon among HLM, more so for leukemias than lymphomas.[2],[4],[5] Among lymphomas, a substantial minority (20%–30%) are complicated by pleural effusion.[6],[7] Involvement of peritoneal cavity and pericardium are far more uncommon, each with a reported prevalence of only 0.8%.[6],[8]
Up to two-thirds of all lymphomatous effusions are said to occur as a primary presentation in a study by Dunphy.[9] Of course, these are the situations that interest cytopathologists as the onus of a credible initial diagnosis depends on them. Effusion cytology as an assessment tool of lymphomatous effusions has a fairly wide detection rate of 14%–88% but is still widely used.[10] This is in view of its ease, personal expertise, and the new ancillary techniques available as part of the laboratory armamentarium.
Over a decade, we came across a good number of HLM diagnosed initially on body fluids, often entirely unsuspected by treating physicians. Our study was intended to assess the concordance with histopathology while having to diagnose these cases on cytomorphology alone. As a by-product, we learnt valuable lessons on restraint, an important virtue when basing a diagnosis purely on morphology.
Materials and Methods This is a retrospective study on stored archival smears of body fluids made on glass slides. In view of the pure retrospective nature of the study with no new tests/procedures being performed and patient identity not being compromised, ethical clearance was waived off by the Institute Review Board. All cases of hematolymphoid malignancies initially reported on fluid cytology over a 10 year (January 2007 to December 2016) period and having follow-up histopathology for correlation were included in the study. The degree of certainty, i.e., whether suspicious or suggestive or confirmatory on the cytology report, was not a bar for inclusion. Cases not satisfying the above inclusion criteria as well as follow-up cases of known HLM were excluded from further analysis.
Eventually, the forms of the cases that satisfy the inclusion criteria were retrieved and the basic demographic details, cytomorphology report, and performance of immunocytochemistry/immunohistochemistry (ICC/IHC) in the cases were noted. Corresponding histopathology reports were also noted.
All the cases where the report was based purely on cytomorphology were individually assigned the most suitable category (discussed below) for further analysis. On the other hand, in all cases where IHC on cell block/ICC on cytologic smears had been performed initially to aid the diagnosis on cytologic material, the morphology slides were blindly reviewed without the performed markers. This was performed by a panel of three experienced cytopathologists who were made aware of only the original clinical details. Eventually, the cytologic category (discussed below) was assigned afresh based on the majority assessment.
For this study, the various reports issued out were condensed into six categories. These were:
Suspicious of hematolymphoid malignancy
This category included cases where there was a strong overall suspicion favoring HLM but morphological difficulties precluded a confident call. Examples would include lineage assessment between leukemic blasts versus blastic lymphomas and doubt between low-grade lymphoma and reactive lymphocytosis.
Non-Hodgkin lymphoma-high-grade (NHL-HG)
This was assigned to cases of unequivocal NHL with high-grade features such as increased mitosis (including abnormal forms), prominent apoptosis, and necrotic background. In this category, lymphoma cells were inevitably large to medium size (larger than or equal to size of histiocyte nucleus) signifying immaturity and further justifying their high-grade behavior.
Non-Hodgkin lymphoma-low grade (NHL-LG)
Cases with monotonous and mature appearing lymphoid cells with low-grade features (lack of mitotic activity, dense nuclei, no necrosis) and minimal, if any, enlargement of the nucleus comprised this category.
Burkitt lymphoma
When the atypical cells showed deep blue cytoplasm with punched out vacuoles and distinct nucleoli with high mitosis and karyorrhectic debris, the case was assigned this category.
Acute leukemia
Atypical cells resembling blasts were assigned this category.
Plasma cell dyscrasia
This was assigned to cases showing unambiguous and frequent presence of either mature or immature plasma cells.
Once the cytologic categorization was complete, reports of cytology and histology were compared for agreement (concordance) or discordance and suitably coded. The discordant slides were retrieved and re-evaluated. A minor discordance was defined as one where there was a therapy unaltering mismatch in morphologic lineage subtyping while agreeing to the basic diagnosis of a HLM. On the other hand, major discordance included those cases where the histopathology (gold standard) did NOT show evidence of HLM or when the provided cytologic diagnosis would lead to an entirely different treatment regimen, despite the cyto-histological agreement on the diagnosis of HLM. Also, when two or more diagnoses were offered, the first differential was taken as the final diagnosis and discordance, if any, was based on that.
Data was tabulated using Microsoft Office Excel Worksheet, version 2007, Redmond, WA, USA. Mean and percentages were calculated using the same software. The discordant group was excluded from all calculations pertaining to epidemiology and cytology of HLM. Free QuickCalcs software from www.graphpad.com accessed on 22nd April, 2020 was used to calculate two-tailed Fisher's exact test for association between the type of effusion and cytomorphologic categories as well as the histopathologic diagnoses. A P value of <0.05 was considered to be statistically significant.
Results Archival search yielded a total of 79 cases where hematolymphoid malignancies were initially diagnosed on fluid cytology over a decade. Among these, diagnostic histopathology reports were available in 54 cases only, satisfying the inclusion criteria. The remaining 25 cases, with unknown histopathology, were excluded from further evaluation.
Among the 54 cases, on comparing cytologic diagnosis with the histopathology, 10 were found to be discordant (discussed in detail later). The results discussed henceforth pertain to the 44 cases with cyto-histo concordance unless explicitly stated otherwise. As per the information on requisition forms, the physicians did not provide a “clinical” differential of HLM in 30 of the 44 concordant cases. Instead, the suspected etiologies included tuberculous etiology, alcoholic liver disease or a simple “effusion for evaluation” without differential diagnosis. At presentation, 63.6%, 25%, and 15.9% of the concordant cases had lymphadenopathy, hepatosplenomegaly, and other forms of organomegaly, respectively.
Males comprised a majority (31 cases, 70.5%) of the HLM cases, with a mean age of 35.7 years (range: 3–72 years). The thirteen females (29.5%) had a comparable mean age of 38.5 years (range: 3–67 years). Both the genders had the maximum cases in the seventh decade of their lives with males having an additional peak in the first decade. This was because of the high numbers of Burkitt and lymphoblastic lymphomas in boys.
Pediatric patients (<18 years age) comprised 29.5% (13 cases) of the cases. On histology, they encompassed eight cases of Burkitt lymphoma, two cases each of T-lymphoblastic lymphoma and T-acute lymphoblastic leukemia, and one case of an uncharacterized B-NHL.
In all, the concordant cases in the study comprised 24 patients (54.5%) of HLM initially diagnosed in pleural effusion, 18 (41%) in ascites, and 2 (4.5%) in pericardial effusion. On histopathology, some interesting observations were made with regards to the afflicted cavity. For example, lymphoblastic lymphomas were almost always noted in pleural fluids. Diffuse large B-cell lymphoma (DLBCL) related effusions were similarly seen to more often manifest as pleural effusion (62.5%). Peritoneum was mostly affected in Burkitt-related effusion (77.8%) and never with lymphoid leukemias, acute or chronic notwithstanding. One case each of myelomatous effusion was seen in the pleural and peritoneal cavities. The rest of the entities were too heterogenous and few in numbers to be subjected to any meaningful scrutiny. Notably, none among the abovementioned observations were associated with a statistical significance but for Burkitt lymphoma (BL) with ascites (P-value = 0.012).
As explained before, after reviewing all reports (with histopathology correlation), we condensed them into six categories. The cases exhibiting concordance between cytopathology and histopathology are illustrated in [Table 1]. Half of the cases (50%) fell into the NHL-HG, i.e., Category 2. Category 1, the proverbial waste-basket category was the next most common category at 13.6% of the cases. BL was a close third comprising 11.4% of the concordant cases and was seen mostly in children and teenagers. NHL-LG, acute leukemias, and plasma cell neoplasms were each represented by less than a tenth of the cases included. There was no statistical association between any of the cytomorphologic categories with the type of effusion (pleural or peritoneal). Notably, Hodgkin lymphoma was not represented in our series.
Table 1: Categories of hematolymphoid malignancies that were newly diagnosed on effusion smears and confirmed on histopathologyConcordance between cytology (purely morphology) and histopathology report (morphology backed by IHC) was seen in 81.5% of the cases (44 out of 54 patients). Major discordance comprised 90% (nine cases) of all discordance and 16.7% (9/54 total cases) of all cases.
In relation to the cytomorphologic categories, discordance [Table 2] was seen in 42.9% of category 3 (3/7 cases) and 25% of the cases comprising categories 1 (2/8 cases) and 6 (1/4 cases). Category 2, the only other category that showed cyto-histo discordance, did so in less than one-fifth (15.4%; 4/26 cases) of its representatives. In absolute numbers, categories 2 (NHL-HG) and 3 (NHL-LG) contributed an alarming 33.3% (3/9 cases) each of major discordance with category 1 (22.2%;2/9 cases) being the next highest contributor.
Table 2: Characteristics of the discordant hematolymphoid malignancies newly diagnosed on effusion smearsFour cases each (44.5% each) of the nine with major discordance were ascertained to have tuberculous lesion and pre-malignant or malignant epithelial lesions elsewhere. One case (11%) represented a wrong call between different lineages of hematolymphoid malignancies. The only minor discordant cases resulted from an upstaging of follicular lymphoma on cytology. The details of discordant cases are chronicled in [Table 2].
Slide review of smears in the cases eventually diagnosed as Tuberculosis revealed a good number of mature lymphoid cells, many with subtle to obviously indented or cleaved nuclei resembling T-cells. Interspersed among these were actually reactive lymphoid cells, initially confused for atypical lymphoid cells/blasts. These cells had discernible basophilic cytoplasm in a quantity intermediate between lymphoblasts and reactive lymphocytes, and irregular nucleus and prominent single regular nucleolus. A representative case illustrating these findings is shown in [Figure 1] with the patient's marrow biopsy showing reactive interstitial polyclonal (CD3>CD20) lymphocytosis. Immature marker, Tdt, was negative. Adenosine deaminase levels and microbiology reports were not available at the time of diagnosis. Radiology, where available, indicated massive effusion or was not definitive of Koch's etiolology.
Figure 1: (a) A reactive lymphocyte rich effusion mimicking lymphoma (MGG,100 ×). The larger cells (above the black asterisks) often have irregular, notched, or cleaved nucleus and prominent cytoplasm. (b) Reactive bone marrow showing lymphocytosis (black arrow) (H&E,100 ×). Marrow shows positivity for (c) CD20 and (d) CD3. (d, Inset) Tdt is negative. (IHC,100 ×)The false-positive cytologic diagnosis in cases of epithelial malignancies was re-confirmed by our panel of three experienced cytopathologists. Uninformed of the discordance, the morphology of the atypical cells was unanimously determined to be compatible with HLM. Barring labeling error, the possibility of a double malignancy was agreed upon. Endocervical curetting showing endocervical adenocarcinoma with positive mucicarmine stain and CEA IHC along with ER negativity and the effusion cytology are shown in [Figure 2].
Figure 2: (a) Large, discrete cells compatible with a high-grade lymphoma in the ascitic fluid. (MGG stain, 100 ×). (b) Endocervical adenocarcinoma (H&E,100 ×) in the same patient confirmed with (c) Mucicarmine stain showing rose pink mucin, (d) CEA positivity, and (d, Inset) negative ER. (400 ×, IHC)The most intriguing misdiagnosis [Case 9 of [Table 2]] pertained to the misinterpretation of reactive plasma cells as myelomatous cells while a small population of myeloid blasts were overlooked as plasmablasts [Figure 3]a. But, timely feedback from hematopathologists regarding the immunohistochemically proven diagnosis of acute myelomonocytic leukemia on peripheral blood smear and bone marrow lineage, forced a slide review. The blasts in the marrow biopsy were positive for MPO (myeloid lineage) and CD14 (monocytoid lineage), while the plasma cells were highlighted with CD138. CD56 used to identify neoplastic myelomatous plasma cells was negative establishing reactive plasmacytosis [Figure 3]b, [Figure 3]c, [Figure 3]d. After concurring on the small numbers of myeloid blasts, a final report of leukemic effusion had been issued from cytology. When the slides were blind reviewed for the purpose of this study, the same mistake was repeated, i.e., the effusion was deemed myelomatous. To corroborate with the hematopathology report, neither the clinical criteria for myeloma were met nor was there any light chain restriction. All this supported myeloid leukemia as the only pathology.
Figure 3: Myeloid blasts (black arrow) amidst mature (black asterisk) and immature plasma cells (white asterisk). (MGG stain, 100 ×). This case was initially misdiagnosed as myelomatous ascites. (b) Marrow biopsy was rich in blasts (black arrow) and plasma cells (white arrow) (H&E,100 ×). Blasts are positive for (c) MPO and CD14 (Inset, c) and plasma cells were (d) CD138 positive and CD56 negative (Inset, d) establishing reactive plasmacytosis in a myelomonocytic leukemia Discussion Hematological malignancies have varied presentation such as fever, B symptoms, generalized/localized lymphadenopathy, and involvement/infiltration of other organs. Involvement of body cavities by HLM may occur as effusions due to multiple mechanisms such as serosal infiltration with cellular shedding, blocked lymphatic drainage through thoracic duct due to external compression from enlarged nodes, excess leucocytosis with cellular exudation, hemorrhage into peritoneum from platelet dysfunction or leukostasis, and extramedullary hematopoiesis (in chronic myeloid leukemias).[10],[11],[12] While NHLs fairly frequently manifest with pleural effusions at some point of the disease course with values quoted from 6% to as much as 48%, the perspective is different for a cytopathologist.[6],[8],[13],[14] Effusions of hematolymphoid origin are rare and comprise a miserly 0.58%–5.5% of all effusions received by a cytology laboratory with an overall average of 0.77% as ascertained by us while collating data from 46,922 effusions from literature.[1],[2],[3],[15],[16],[17],[18] Among “malignant” effusions, few large studies provide a prevalence of 10%–17.1% for hematolymphoid etiology.[1],[2],[14],[19]
Given the rarity of HLM manifesting as effusions, the chance of encountering the same as first presentation is rarer and provides an even more unique opportunity for the cytopathologist. To the best of our knowledge, no such compilation of newly detected HLM has been attempted and is the novelty of our work.
Males comprised a majority of the included cases in our study as was also seen in some other articles.[1],[2],[6],[19] However, unlike previous studies, our pediatric age population (29.5%) contributed far less than prevalent data that ranged from 52% to 81%.[20],[21] Being prone to high-grade lymphomas, pediatric hematolymphoid effusions are seldom misleading. In fact, concordance in that population was about 92% in this study with only one exception. Thus, while the reduced frequency appears to be a credible finding, we are unable to provide a reason for the same.
When compared to the literature, the relative involvement of pleura, peritoneum, and pericardium in HLM in our series was less than, more than, and within the ascertained range of 4.2%–10.3%, 16.7%–40%, and 60%–79.1%, respectively.[2],[3],[15],[16],[22],[23]
Another point to be noted is the relative preponderance of a BL diagnosis in our series over the conventional lymphoblastic leukemia/lymphoma entity.[16] It may have to do with ours being a tropical country with an increased risk of Epstein Barr virus infection. The common HLM in our series were high-grade B-NHL such as BL, DLBCL, and LBL, which is only in partial agreement with Lew et al.[16] They uniformly illustrated good numbers of DLBCL, a significant paucity of Burkitt effusion (even in ascites) and a total absence of lymphoblastic lymphomas in the 58 HLM detected in their series from Michigan, United States. Presumably, early presentation of lymphoblastic lymphomas due to awareness arising from higher socioeconomic status and paucity of EBV infections in non-tropical settings may explain the disparity in numbers. There were no cases of Hodgkin lymphoma in our series, though an occasional example is rarely cited.[24]
For this study, the various reports issued out were condensed under six categories, which were mainly influenced by therapeutic implication but moderated with morphologic consideration and the overall potential ease of use. This would explain the need to classify Burkitt lymphoma and plasma cell dyscrasias separately. In view of their unique cytomorphology and treatment, merging them under the umbrella term of NHL-HG would not be appropriate on a cytology report and not benefit a patient. While categorization appears to not be difficult, this study does expose limitations in the diagnosis of HLM purely on cytomorphology. Tuberculosis and epithelial malignancies contributed to the majority of these situations in our series.
Lung disease secondary to histology proven tuberculosis may be easily confused with NHL-LG especially when clinical or radiological assessment is obscured by a massive effusion. The sensitivity of a cytomorphologic diagnosis is stated to be a dismal 10% when appropriate clinical details and ancillary studies are not available.[25],[26] Overconfidence on scarce suspicious cells should be avoided because they are more likely to represent a reactive process. Rather, it is best the effusion cytology report be withheld till further biochemical, radiological, histopathological, or microbiological reports are available. At the time of this study, flow cytometry and immunocytochemistry were not available in our department, which definitely contributed to the misdiagnosis of reactive effusion as neoplastic effusion on morphology. Differentiating reactive lymphocyte from lymphoma cells with naked eye can be incredibly difficult. Morphometry is said to help solve the conundrum somewhat with a high, yet not perfect, specificity of >95% but is not easy to perform.[6],[27],[28] A word of caution on immunocytochemistry. One must be aware that ICC for B- and T-lymphocytes to look for B-predominance (as a marker of a clonal process) may not be foolproof as T-cell rich effusions in chronic lymphocytic leukemia/small lymphocytic lymphoma, the classic B-cell neoplasm, are also known.[29] Additionally, the predominant lymphocyte population in any effusion, including tuberculous effusion, are T-cells.[14] As a corollary, B-cell rich effusions are likely to be abnormal.
The false positive cytologic diagnosis in our cases of epithelial malignancies could be explained as confusion from mislabeled slides or genuine double malignancy such as chemotherapy-induced secondary NHL. Lack of epithelial fragments may not be always helpful because neoplasms such as amelanotic melanoma, some nasopharyngeal carcinomas, and poorly differentiated carcinomas are discohesive.[11] Large reactive lymphocytes such as Downey cells ascribed to infectious mononucleosis are another potential pitfall.[11]
The myeloma-leukemia discordance continues to bewilder us. However, the mere knowledge that myelomatous effusions are as such rare and seen in <1% of myelomas, should make anyone skeptical while reporting them unless unequivocal supporting evidence exists.[13],[30],[31] Myelomatous ascites (as was our case) is even rarer with not a single case in a series of 30,085 malignant effusions.[2]
Other potential for overdiagnosis includes interpreting leukemic effusion on hemorrhagic aspirates. In such situations, macrophages and mesothelial cells must be actively searched for. Communication with the treating physician and procuring meaningful samples such as a finger prick smear in suspicious cases are crucial.
On a positive note, an 81.5% cyto-histo concordance is commendable given the fact that these HLM presented initially as effusion and were reported out purely on cytomorphology, often without any pointing history. This augurs well for many cytopathologists working in small labs without recourse to advanced techniques or even second opinion.
Conclusion Using this nomenclature, satisfactory concordance was observed between cytomorphology and histopathology on a difficult morphologic domain. Yet, our experience suggests an upfront diagnosis of HLM entities on cytology may be restricted only to cases with an unequivocally abundant atypical population. Reactive lymphocytes are close and recurrent mimics of atypical cells and additional ancillary tests should be resorted to at the slightest doubt. Fortunately, appropriate ancillary techniques also require a good morphologic assessment whose importance, therefore, is still paramount.
Acknowledgements
The first author would like to thank Mr. Sushil Patel, Sri Aurobindo Ashram, Puducherry for his help in editing the photographs to meet the publisher's standards. All authors would like to thank the various histo- and hematopathologists involved in reporting the study cases over the decade.
Financial support and sponsorship
Nil.
Conflicts of interest
There are no conflicts of interest.
References 
Source of Support: None, Conflict of Interest: None
CheckDOI: 10.4103/joc.joc_204_21
留言 (0)