Destaining hematoxylin and eosin stains and restaining for immunohistochemistry has diagnostic value for cytology samples

INTRODUCTION

Fine needle aspiration (FNA) is a minimally invasive technique with multiple advantages, including a lower risk of procedure-related complications such as bleeding or infection, better preservation of cytomorphology, minimal requirements for anesthesia, sampling different areas of a lesion with one pass, rapid on-site diagnosis, and the ability to provide material for multiple types of ancillary studies. The addition of a cell block preparation allows for the performance of ancillary tests that significantly enhance the diagnostic accuracy of cytopathologic examination, including immunohistochemical (IHC) and molecular studies.^[1] However, FNA cytopathologic specimen examination has its limitations, the major concern being the relative lack of tumor architectural pattern when a only a small amount of cells is aspirated and examined, which in turn leads to a haphazard distribution of neoplastic cells, especially when sparse and admixed with a rich background of other benign and inflammatory cell populations, thereby making it often challenging to track specific target cells at different levels on IHC slides, precluding accurate interpretation of IHC results.

This is especially an issue when it comes to the diagnosis of hematopoietic disease on cytology specimens, which is often based on the IHC expression pattern of the neoplastic cells that often have a rich background composed of other cell populations, and the need for multiple IHC biomarkers for accurate subclassification of hematopoietic diseases.^[2] The routine process of IHC staining, one IHC marker per level, creates even more diagnostic challenges due to the difficulty in tracking the scant neoplastic cells at different levels and the risk of losing valuable cellular material with each subsequent slide cut. In addition, since some lymphomas require architectural assessment for diagnosis and classification, it can be challenging for hematopathologists to make a diagnosis based solely on FNA cytology specimens, and the current lymphoma guidelines recommend against using FNA cytology alone for the primary diagnosis of lymphoma.^[3-5] As a result, patients often have to go through subsequent, more invasive excisional or core needle biopsies, which can result in treatment delays or procedure-related complications.

Pathologists have endeavored to improve the diagnostic accuracy of lymphoma on FNA specimens. For example, Zeppa et al. recommend combined flow cytometry (FCM) and cytology analysis of FNA materials^[6,7] as FCM is an especially helpful tool in the detection and classification of small cell lymphoma, while cytology can often detect large, atypical cells in large cell lymphomas, including diffuse large B-cell lymphoma (DLBCL) and Hodgkin’s lymphoma (HL). Their large-scale data analysis, including 245 lymphoma cases and 188 benign lymph node cases, showed that combined FNA and FCM have a high sensitivity (94.9%) and specificity (99.4%) in the diagnosis of non-HL (NHL) with more consistent and reproducible results across different laboratories. Limitations of this technique of combining cytology and FCM were mostly related to sampling error due to lymphoma partially involving a lymph node, lowcellularity, cell loss during processing, and FCM issues, such as clonality not being performed in rare T-cell NHL. The only false-positive case was caused by chemotherapy-related reactive cytologic changes in which the FCM on the FNA specimen showed CD10/CD19 co-expression with λ light-chain, while surgical excision of the lymph node demonstrated a florid follicular hyperplasia.^[8] Although combining FNA and FCM is a promising methodology in the diagnosis of NHL on cytology specimens, the improvement of the technique can be limited by the scant nature of the specimen, along with other technical and clinical challenges.^[6] Other groups have explored the use of FNA samples for molecular studies, including fluorescent in situ hybridization, polymerase chain reaction (PCR)-based assays, and gene sequencing techniques.^[9,10] However, they may face similar issues as other techniques, mainly due to the sparse cellularity and difficulty in tracking and following the neoplastic cells from the cytomorphologic examination stage to the ancillary studies’ stage.

We explore the techniques to solve the above problem, which we call the “cell tracking technique.” In this study, we applied a well-developed method Destaining hematoxylin and eosin (H&E) stains and restaining (DSRS), to track the expression of IHC biomarkers of the same individual lesional cell among a large background of other non-lesional cell populations, with the help of image scanning. The goal of the present study is to assess the utility of DSRS on cytology cell blocks as a possible cell tracking technique for accurate disease diagnosis.

MATERIAL AND METHODS

Case selection

Institutional Review Board approval (STUDY10396 and STUDY2004) from UMass Chan Medical School was obtained for this retrospective study, and patient informed consent was waived. This pilot study includes cases with cell tracking issues selected from our daily practice, spanning from December, 2016 to November, 2021. The inclusion criteria are as follows: (1) Initial diagnosis on cytology FNA sampling was “suspicious for neoplasm/malignancy” with difficulty in interpreting critical IHC markers for diagnosis or treatment. (2) The cell block H&E slides revealed a rich background of mixed benign epithelial and inflammatory cells, with rare scattered atypical/suspicious cells arranged either singly or in small clusters [Figure 1a]. Cell blocks lacking atypical cells will be excluded from the study.

Close

Figure 1:

Case 1, “para-esophageal” mass fine needle aspiration (FNA). (a) cell block, H&E stain. A relatively sparse abnormal population of cells of possible hematopoietic origin is present. The cells are seen singly in a background of numerous histiocytes, lymphocytes, mesothelial cells, and pneumocytes. Large atypical cells are seen with large folded nuclei with open chromatin and scant cytoplasm. (b) Same case after DSRS with CD79a IHC stain, highlighting scattered singly large atypical cells. This positive stain, along with the cytomorphology, was highly suspicious for a high-grade B-cell lymphoma on cytology FNA specimens. Scale bar: 0.05 mm. 20× magnification (total 200×)

Export to PPT

The cellular material obtained from FNA sampling was collected in CytoRich red and then made into a ThinPrep and a Cellient cell block. The H&E slides were scanned using GT450 (Leica Aperio GT450, Leica Biosystems, Deer Park, IL), destained, and then restained with one IHC biomarker per H&E slide [Figure 1a]. The subsequent IHC-stained slides were also scanned, and the images were saved for evaluation in e-slide manager software [Figure 1b]. The software allows the synchronization (by manual alignment) of up to four whole-slide images from the same case, allowing for accurate tracking of limited atypical or suspicious neoplastic cells of interest, and accurate interpretation of IHC results of individual target cells [Figure 2]. The IHC antibody information has been listed in Table 1. IHC results were validated for the DSRS process for all antibodies without modification of the IHC protocol.

Close

Figure 2:

DSRS synchronized 4-panel view, Case 4. (a and b) H&E stain (a) and corresponding CD30 IHC stain (b) after DSRS. H&E stain demonstrates typical Reed-Sternberg (RS) cells with large nuclei and prominent nucleoli; CD30 immunostain shows strong membranous and Golgi-zone positivity in the same RS cells (arrows). (c and d) Serial section with H&E stain (c) and corresponding CD15 IHC stain (d) after DSRS, highlighting the CD15-positive RS cells (arrows). This perfect pixel-level registration allows direct side-by-side comparison of the morphology and immunoprofile of the same individual tumor cells using DSRS, which is particularly useful in classical HL, where RS cells can be sparse. Scale bar: 25 μm. 40× magnification (total 400×)

Export to PPT

Table 1: Immunohistochemical staining antibody information.

Antibody	Maker and catalog #	Clone	Concentration	Retrieval
CD30	Dako M0751	BerH2	1:40	Citrate pH 6.0
CD15	B&D 347420	LeuM1	1:20	Citrate pH 6.0
MUM1	Dako M7259	MUM1p	1:2000	Citrate pH 6.0
PAX5	Cell Signaling 12709	D7H5X	1:50	Citrate pH 6.0
Synaptophysin	Dako M0776	SY38	1:160	Ethylenediaminetetraacetic acid pH 8.0
Chromogranin	Dako M0869	DAK-A3	1:8000	Citrate pH 6.0
Ki67	Dako M7240	MIB-1	1:100	Citrate pH 6.0
CKAE1/3 (PanCK)	Dako M3515	AE1/AE3	1:100	Citrate pH 6.0
BOB1	Leica NCL-L-BOB-1	TG14	1:80	Citrate pH 6.0
CD20	Dako M0755	L26	1:600	Citrate pH 6.0
OCT2 CD79a	Leica NCL-OCT-2 Roche 790-4432	OCT-27 SP18	1:50 No dilute	Citrate pH 6.0 Roche’s CC1

RESULTS

To evaluate the sensitivity of the DSRS technique, nine cases were chosen. The selected FNA cases were primarily from lymph nodes (cervical, retroperitoneal, and inguinal) and the lung. One case, initially identified by radiology as a left para-esophageal mass, was later confirmed during surgery to originate from the left lower lung lobe [Table 2]. Patient ages ranged from 18 to 75 years, including six females and three males. The final diagnoses were as follows: Six cases of classical Hodgkin’s lymphoma (cHL), one case of nodular lymphocyte-predominant Hodgkin’s lymphoma (NLPHL), one case of CD20-negative DLBCL, and one case of lung carcinoid with a low Ki67 proliferation index (<1%) [Table 2].

All cases were independently reviewed by two blinded observers (M.A. and T.Z.), with full inter-observer agreement. Cases with any question or discrepancy from the confirmatory result were also reviewed by Dr. SK (hematopathologist). The diagnoses made on the limited FNA samples, both before and after applying the DSRS method, were compared with the confirmatory diagnoses rendered on specimens containing more abundant cellular material (gold standard), including core biopsies, resection/excision specimens, or additional cytologic samples with abundant cellular material. A true-positive neoplasm diagnosis is defined by the expression of tumor markers (e.g., co-expression of CD30 and CD15 for cHL) with correlation to neoplastic cells on the H&E image.

In all the cases included in the study, the DSRS technique was essential for achieving a definitive diagnosis or determining tumor differentiation. For example, case 1 was a 75-year-old woman with a history of breast cancer who presented with worsening back pain. Imaging showed a T7 compression fracture, and FNA of the bone lesion was inconclusive due to sparse atypical cells. Chest CT reviewed by senior pathologist AHF revealed a 3.6 cm “para-esophageal” mass, prompting referral for FNA. On-site cytology evaluation showed numerous large atypical cells, but the cell block [Figure 1a] contained only a few atypical cells amid numerous histiocytes, lymphocytes, mesothelial cells, and pneumocytes, complicating tumor cell tracking across IHC levels. Despite extensive workup, the tumor cells seemed to be negative for most lineage markers, including carcinoma, lymphoma, sarcoma, and melanoma. The case was initially signed out as “Highly suspicious for malignancy” after hematopathology and soft tissue consultation. DSRS applied to an H&E slide [Figure 1b] identified CD79a-positive large neoplastic cells, suggesting a rare CD20-negative DLBCL. Surgery clarified that the mass was actually a left lower lobe lung mass, and wedge resection confirmed the diagnosis.

In our study, three pathologists reached full concordance in IHC interpretation on all the study cases. Immunostain interpretation on DSRS was 100% concordant with IHC on more adequate specimens, although not all cases yielded a definitive diagnosis. Case 7 was the only case without a confirmed diagnosis post-DSRS, due to the absence of CD15 expression in Reed-Sternberg (RS) cells on the cytology cell block. The follow-up cervical lymph node excision confirmed a diagnosis of cHL, despite the RS cells being CD15-negative. The diagnosis was supported by the presence of classic RS cells and their immunoreactivity for CD30, PAX5, and BOB1, along with diminished expression of other B-cell markers, including negative staining for OCT2 and weak, subset positivity for CD20 and CD79a. This immunophenotypic profile, together with characteristic cytomorphology, led to the diagnosis of cHL. As a result, the CD15-negative atypical RS cells in the cytology specimen were concordant with the IHC profile observed in the subsequent surgical specimen.

Our study showed that the DSRS technique on limited FNA samples did not compromise tissue quality or IHC integrity. Comparison of the limited FNA diagnosis and final diagnosis rendered on more adequate specimens revealed concordant and accurate results in all cases, except in one case where the absence of definitive CD15 staining in tumor cells prevented a confirmed diagnosis of cHL on a limited cytology specimen [Table 2, case 7]. The diagnostic concordance rate before the application of DSRS was 22% (2/9), and the concordance rate after the application of DSRS was 89% (8/9), while the immunostain interpretation concordance rate was 100% (9/9). Our pilot study showed that the DSRS technique effectively overcomes the challenges of limited material in cytology specimens.

DISCUSSION

DSRS of cytology cell block slides offers a promising technique to potentially overcome the challenge of tracking sparse cells of interest in a benign cellular background, especially in cases where a panel of IHC stains is required to achieve a diagnosis. H&E staining is the most widely used method for tissue diagnosis. The darker purple color is produced by the dye hematoxylin, which stains basophilic cell structures such as nuclei, while the pink color is generated by the acidic dye eosin, staining other cell compartments, including cytoplasm, and tissues such as stroma and muscle fibers.^[11-13] One advantage of H&E staining is that it allows for the destaining of the slide to be used again for another IHC stain, especially with the help of whole-slide imaging that preserves the initial morphology before destaining the original H&E slides.^[13]

Although DSRS itself is not a new technique, its application to cytology cell blocks has not been reported, nor has this technique been used to address the issue of cell tracking on a limited FNA sample or small biopsy specimens. The digitalization of images has made it possible to simultaneously compare the cytomorphology of neoplastic cells and the expression of IHC biomarkers. As residual H&E is cleared by routine IHC steps (ethanol series, buffer washes, and antigen retrieval), the procedure poses no or minimal additional risk of antigen degradation. DSRS technique provides a more reliable and cost-effective solution compared to newer, more costly techniques such as multiplex IHC/IF. However, the DSRS technique has several potential limitations: Questionable reliability of the GT450 scanner for digitizing cytology slides, restriction to one IHC marker per slide, and the time-consuming workflow.

More advanced techniques to overcome the cell tracking issue include Cocktail IHC, chromogenic multiplex immunohistochemistry (mIHC), multiplex immunofluorescence (mIF), and tissue-based mass spectrometry. Below, we will briefly discuss the advantages and shortcomings of each of these techniques.

1. Cocktail IHC: Currently, the most widely used chromogens for detection are hydrogen peroxide oxidized DAB, producing a brown color, followed by 3-Amino-9-ethylcarbazole, producing a red color. The combination of the two chromogens, along with the different distribution of the antibodies (nuclear vs. cytoplasmic), allows for up to three-antibody staining on a single slide. One well-known cocktail, IHC, is PIN4 (P504S, P63, high molecular weight cytokeratin), widely used in prostate cancer diagnosis.^[14] Recently, ADH-5 (CK5/14, p63, CK7/18 antibodies - Biocare Medical, Concord, CA) has been used for the diagnosis of atypical ductal hyperplasia, usual ductal hyperplasia, and in the evaluation of invasion.^[15,16] Similar to traditional chromogenic IHC assays, this technique has well-defined protocols and sets of instructions that are relatively easy to follow. Light microscopy reading can be captured by brightfield digital image scanners and is inexpensive. However, it cannot address the cell-tracking issue if the antibody distributions (nuclear vs. cytoplasmic) are similar, or if more than 3 antibodies are needed to stain a single cell.

2. Multiplexing techniques allow the analysis of a single cell with multiple antibodies (2-50) conjugated with either chromogens or fluorophores (mIF), by applying a stain and strip technique or cycling approach, respectively. MIHC is made possible by the emergence of new chromogens with multiple colors, including purple, red, and blue, among others.^[17] MIHC requires the chromogen (enzyme substrate) to be soluble in chemicals (such as alcohol) so that the chromogen can be stripped before another cycle of staining with a new antibody is performed, followed by image scanning for each single chromogen staining. Theoretically, it is an attractive technique to solve the cell tracking issue with obvious advantages, including multiplexing with up to 10 markers on a single slide, rendering a multiplex whole slide image, as well as allowing for the use of brightfield microscope reading and brightfield digital image acquisition platforms, which most pathologists prefer. Shortcomings of this technique include that it is time-consuming (1 day per cycle and 10 days for ten markers), the possibility of tissue loss during coverslip removal, and the requirement for labor-intensive validation, which affects the popularity of the technique. In addition, it is usually difficult to multiplex with chromogen, as currently only a few chromogens can demonstrate marker co-expression effectively on a single cell.^[18] Therefore, most labs using the mIHC technique do not include the step of individual slide scanning, but instead scan one composite image, which is helpful in cases where the biomarkers of interest are located on different cells, but not useful for detecting one cell with co-expression of multiple markers.

3. Multiplex IF: Multiple platforms exist for mIF techniques, including standard IF scopes, which can support 4–5 markers, and multispectral technologies (Vectra 3.0^™/Polaris^™), which can support 6–8 markers. The mechanism is that multiple antibodies with selective fluorophore labels can target different epitopes; the individual fluorophores can be excited to emit a specific wavelength that can be captured by a distinct filter in an in situ fashion, allowing for the study of the co-localization of targets in single or different cells. The tyramide signal amplification technique is widely used in mIF; in this technique, tyramide can covalently bind to the epitope after signal amplification. While the primary/secondary antibody complexes that are noncovalently bound to the epitope can be removed by heat, the tyramide fluorophore that is covalently bound to the tyrosine residues surrounding the epitope remains on the cell.^[17,19] The advantage of mIF is that it allows spatial studies in an individual cell, and the quantitative study of marker intensity, as well as faster turnaround time compared to Mihc.^[14] The potential disadvantage of mIF is that the number of markers may be limited (4–5 colors) due to the false-positive reading by neighboring channels if the fluorophore spectrums are too close in wavelength, while the multispectral imaging system can partially solve this problem (up to 8 colors) by carrying out linear unmixing of signals, which can also correct tissue autofluorescence.^[18,20]

4. Tissue-based mass spectrometry or elemental mass spectrometry immunohistochemistry (MIBI): This technique directly images elemental mass tags on primary antibodies in a single tissue section. Although MIBI is currently limited to research purposes due to high cost, the requirement for intensive training of lab staff, and the long periods of time needed for reagent optimization and imaging, this technique has its advantages: it can simultaneously detect multiple markers and has no autofluorescence issue, thus making it possible to measure over 100 markers and quantify marker intensity.^[19,21]

In clinical practice, cHL is typically characterized by the presence of RS cells within a mixed inflammatory background and a distinctive immunophenotypic profile. RS cells in cHL are usually positive for CD30 and CD15, with weak expression of the B-cell marker PAX5, and negative for CD45 and CD20 in most cases. Therefore, the integration of cell tracking techniques can notably enhance diagnostic accuracy. However, CD15 negativity is observed in a minority of cHL cases, and this does not exclude the diagnosis when supported by characteristic morphology and expression of other markers. The proportion of CD15-negative cHL cases varies across studies. In a comprehensive analysis by the German Hodgkin Study Group involving 1,751 cHL cases, 12% were found to be CD15-negative (CD15-, CD30+, CD20-). Another study from India reported a higher rate, with 35.8% of cHL cases lacking CD15 expression. In addition, a retrospective study indicated that 12–20% of cHL cases were CD15-negative.^[22,23] This underscores the importance of using a comprehensive panel of markers to accurately establish the diagnosis, particularly in atypical or immunophenotypically variant presentations of cHL.

In a small biopsy or a cytology specimen, a definitive diagnosis of cHL requires extreme caution due to the existence of several cytological mimickers of RS cells, including but not limited to immunoblasts in infectious mononucleosis,^[24] reactive lymphocytes in Epstein-Barr virus-positive mucocutaneous ulcers,^[25] and some large cell lymphomas such as mediastinal grey zone lymphoma.^[26] If the pathologist encounters any doubt regarding the specific final diagnosis, it is recommended to provide a differential diagnosis whenever possible and allow the clinician to decide whether to favor one diagnosis over the other based on the clinical presentation or to perform an excisional biopsy for a definitive diagnosis and treatment.

SUMMARY

DSRS provides a cost-effective approach for tracking minute lesional cells in cytology specimens and has proven clinically useful for disease diagnosis using a panel of IHC stains. With technical support from digital cytology platforms such as the Hologic imaging system, the integration of DSRS into routine practice can be further enhanced.

AVAILABILITY OF DATA AND MATERIALS

The datasets and additional materials are available from the corresponding author upon reasonable request.

ABBREVIATIONS

H&E: Hematoxylin and eosin

DSRS: Destaining H&E stains and restaining

FCM: Flow cytometry

HL: Hodgkin’s lymphoma

NHL: Non-HL

cHL: Classical Hodgkin lymphoma,

DLBCL: Diffuse large B-cell lymphoma,

NLPHL: Nodular lymphocyte predominant Hodgkin lymphoma,

RS: Reed-Sternberg

LN: Lymph node

FNA: Fine needle aspiration

IHC: Immunohistochemical

PCR: Polymerase chain reaction

mIHC: Multiplex immunohistochemistry

mIF: Multiplex immunofluorescence

MIBI: Mass spectrometry immunohistochemistry

AUTHOR CONTRIBUTIONS

MA: Data analysis and draft manuscript; TZ: Study design, case collection, data analysis, draft and revise manuscript; SK: Hemepath consultation, revise manuscript; MZ: Study design; KD: Technical support; AHF: Study design. All authors approve of the final manuscript and meet the ICMJE author qualifications.