Role of flow cytometric immunophenotyping in the diagnosis of breast implant-associated anaplastic large cell lymphoma: A six-year, single institution experience

Abstract

Objectives:

Breast implant-associated anaplastic large cell lymphoma (BIA-ALCL) is an uncommon mature T-cell neoplasm occurring in patients with textured breast implants, typically after 7–10 years of exposure. While cytopathologic or histopathologic assessment is considered the gold standard diagnostic method for BIA-ALCL, flow cytometry (FC) based immunophenotyping is recommended as an adjunct test. However, the diagnostic efficacy of FC is not well reported.

Methods:

We reviewed 290 FC tests from breast implant peri-capsular fluid and capsule tissue from 182 patients, including 16 patients with BIA-ALCL over a 6-year period, calculating diagnostic rates and test efficacy.

Results:

FC showed an overall sensitivity of 75.9%, specificity of 100%, and negative and positive predictive values of 95.4% and 100%, respectively. Blinded expert review of false negative cases identified diagnostic pitfalls, improving sensitivity to 96.6%. Fluid samples had better rates of adequate samples for FC testing compared to tissue samples. Paired with FC testing of operating room (OR) acquired fluid samples, capsulectomy FC specimens added no diagnostic value in patients with concurrent fluid samples; no cases had positive capsule FC with negative fluid FC.

Conclusions:

Fluid samples are adequate for FC testing more often than tissue. Capsule tissue FC specimens do not improve FC efficacy when paired with OR-acquired fluid FC samples and are often inadequate samples. FC is 100% specific for BIA-ALCL and can serve as a confirmatory test but should not be the sole diagnostic method. Awareness of sample-specific diagnostic pitfalls greatly improves the sensitivity of BIA-ALCL testing by FC.

1. INTRODUCTION

Breast implant-associated anaplastic large cell lymphoma (BIA-ALCL) is an uncommon mature T-cell lymphoma arising in the proximity of breast implants, first described in 1997 and recognized as a discrete entity by the WHO in 2016.^1–4 It presents with non-specific findings such as breast swelling, inflammation/erythema, or asymmetry.^1,5 Signs and symptoms such as non-resolving seroma (periprosthetic fluid collection), swelling, pain, palpable mass, and tenderness are among the most described, while late-onset effusion resulting in a change in breast size is the most typical presentation.^5–7 Most patients are diagnosed after a median exposure to textured implants of 11 years (ranging from 5 to 15 years). The highly textured surface implants implicated in many studies in the development of BIA-ALCL were withdrawn from the US, EU, and Brazilian market in 2019. The FDA estimated that approximately 550,000 breast implants are placed every year in the US including 70,000 textured devices, though this percentage may have decreased below 5% after the 2019 recall. Approximately 2 million women are estimated to have textured-surface devices in place.^8–12 Diagnosis is typically made by cytologic evaluation of peri-prosthetic fluid collection after radiologic evaluation via ultrasound or magnetic resonance imaging. However, the majority of peri-prosthetic fluid collections are benign.^6,13–17 Evaluation of positive seroma fluid is then followed by implant removal and capsulectomy to assess the degree of capsular infiltration and stage the disease.^2,6,14,18

Flow cytometric (FC) evaluation is essential in diagnosing most hematopoietic neoplasms, allowing rapid assessment to guide morphologic and immunohistochemical (IHC) evaluation. The immunophenotype of BIA-ALCL has been described mostly through IHC study. It is essentially indistinguishable from other ALK-negative ALCLs, with strong, diffuse positivity for CD30 and frequent loss of surface pan-T-cell antigens such as CD3, CD5, and CD7. They may retain CD2 and CD4, and often also express CD25 and HLA-DR.^19–21 Flow cytometry studies performed early in the patient work-up on fine needle aspiration (FNA) fluid can provide a confirmation of a cytology diagnosis and help guide further management. Flow cytometry studies performed on fluid acquired in the operating room (OR) during seroma drainage can also provide confirmation. Flow cytometry performed on fluid is especially useful if the disease is confined to the capsular fluid or has only focal capsular involvement. While FC is recommended as an adjunct test for the workup of BIA-ALCL by the National Comprehensive Cancer Network (NCCN), the diagnostic utility of FC studies in diagnosing BIA-ALCL has not been systematically assessed from an efficacy standpoint.¹⁴

Various samples are sent for pathologic evaluation for BIA-ALCL, especially fluid samples acquired through cytology procedures such as FNA, fluid obtained in surgical procedures performed in an OR setting, and capsulectomy specimens. The value of FC in diagnosis may vary between specimen types.⁶ In addition, patients may undergo multiple drainage procedures for symptomatic relief, which can affect test efficacy by reducing the number of diagnostic neoplastic cells.⁶ Flow cytometry offers unique properties in terms of rapid diagnosis and is recommended by the NCCN an adjunct technique to supplement the morphologic diagnosis of BIA-ALCL, though the efficacy of FC is not well documented. In this study, we present a 6-year, single-center experience in a large tertiary care cancer center using FC in routine pathologic evaluations for BIA-ALCL, assessing its efficacy and diagnostic utility.

2. MATERIALS AND METHODS

2.1. Sample selection

The study was approved by the Institutional Review Board. Pathology archives were searched from 2016 through 2021 for all FC samples from breast implant capsular tissue or breast implant adjacent fluid. Cytologic, histopathologic, and FC results were reviewed. For the purposes of calculating test efficacy, the cytologic or histopathologic diagnosis was considered the “gold standard” for the diagnosis of BIA-ALCL; samples with a final cytology or histopathology diagnosis consistent with BIA-ALCL were considered positive for disease, and samples with both cytology and histopathology final diagnoses other than BIA-ALCL were considered negative.

2.2. Flow cytometry sample preparation

Flow cytometry was performed on all samples as part of routine clinical practice at a major tertiary cancer center. Tissue biopsy specimens were prepared in the standard manner used for routine clinical FC; tissue was finely minced in RPMI 1640 growth medium using scalpel blades, then filtered and resuspended in RPMI. Breast implant capsule sections were measured in 3 dimensions, inked on the outer surfaces, carefully examined for nodules or masses, and then serially sectioned. If a nodule, mass, or other abnormality was detected, a sample (approximately 0.5 cm³) was submitted for FC testing and processed as a tissue biopsy specimen as above; if no gross abnormality was detected, a random sample of similar size was submitted for FC testing. Needle aspirates and biopsy needle rinses were filtered and suspended in RPMI. A rough total sample cellularity calculation was performed using a Cellometer device (Nexcelom Bioscience, Massachusetts, USA) to measure cell concentration, then multiplying this concentration by the volume of suspended sample, as previously described²².

Following resuspension, cells were stained with a 10-antibody panel designed for evaluating potential CD30 positive T-cells using antibodies against CD2 (BD Biosciences [BD], New Jersey, USA), CD3 (BD), CD4 (BD), CD5 (BD), CD7 (Beckman Coulter [BC], California, USA), CD8 (BD), CD30 (BC), CD45 (BD), CD56 (BC), and CD279 / PD-1 (Biolegend, California, USA). A subset of cases also included staining with 4′,6-diamidino-2-phenylindole (DAPI) to assess cell viability. Following staining and incubation, cells were lysed, fixed, washed, and resuspended. Up to 500,000 cells were acquired on flow cytometers including BD Canto 10 color flow cytometers, BD Fortessa 18 color flow cytometers, and BD Symphony 26 color flow cytometers (BD). The use of multiple different flow cytometer models was due to routine laboratory equipment changes and improvements, and new instruments were validated with approval by the New York State Department of Health. Flow cytometry panels are summarized in Supplemental Table 1.

2.3. Flow cytometry immunophenotypic assessment

Results were analyzed with custom Woodlist software (generous gift of Wood, B.L., University of Washington, Seattle, WA, USA). All assays were initially evaluated by a senior technologist and a single pathologist. Abnormal CD30-positive T-cell populations were identified by visual assessment of populations with antigen expression profiles fitting previously reported phenotypes of BIA-ALCL (Figure 1).^19–21,23 A mononuclear cell gate was used to search for CD30 positive cell populations. The mononuclear cell population was gated using a series of sequential gates for singlet cells, followed by viable cells, then mononuclear cells, as previously described²². A sample was considered “adequate” if it met 1 of the following conditions: (1) an abnormal population (10 or more cells) was detected by FC; or (2) at least 10,000 viable cells were acquired for analysis, based lack of DAPI staining by FC, or by forward scatter (FSC) and side scatter (SSC) gating strategies, whichever method gave the lowest cellularity counts. Samples were considered inadequate if both of the following conditions were met: (1) fewer than 10,000 viable cells were acquired for analysis; and (2) no abnormal population was detected. Samples with fewer than 10,000 viable cells were still analyzed as the samples are considered irreplaceable; however, they were reported with a disclaimer stating that the samples were of suboptimal quality. However, these samples were omitted from further efficacy calculations. The FC assay can detect abnormal populations (10 or more cells) down to 0.1% of viable cells. Rare cases which were reported as “atypical” for unusual immunophenotypic features were reviewed separately. As all cases in this study were submitted for routine clinical testing, evaluating pathologists were not formally blinded to cytologic or histopathologic results, but FC specimens were usually reviewed prior to cytology or histopathology specimens.

Figure 1.

Example of flow cytometric findings in BIA-ALCL. A. Flow cytometry of peri-implant seroma fluid shows a surface CD3 negative, CD30 bright population (red population). Note that the CD30 expression is bright and uniform. B. The population expresses CD2, and most of the population shows a loss of CD7. C. CD4 is positive in this population, slightly dimmer than background CD4 positive T-cells, while CD8 is negative. D. The population also demonstrates markedly increased forward and side scatter, suggestive of large size and cellular complexity.

Abbreviations: BIA-ALCL – Breast implant associated anaplastic large cell lymphoma.

2.4. Statistical analysis

Statistical testing was performed using Prism (GraphPad, San Diego, CA, USA). Fisher’s exact test was used for comparisons between categorical variables. Sensitivity, specificity, positive predictive value (PPV) and negative predictive values (NPV) were also calculated, alongside the Receiver-Operating Characteristic (ROC) curve. Findings were considered statistically significant for P-values of < 0.05.

3. RESULTS

3.1. Clinicopathologic Features

Clinicopathologic features are summarized in Table 1. From 2016 through 2021, we identified 290 FC samples assessing for BIA-ALCL from 182 unique female patients. 119 of 182 (65.4%) patients had at least 1 FC sample deemed adequate for FC diagnosis. Sixteen patients with BIA-ALCL were identified, of which 15 (93.8%) had at least one positive FC test. As some patients had multiple FC tests performed over the course of their management, there were 22 total samples that were positive for BIA-ALCL by FC. For patients where this data was available, there was 1 patient with a smooth implant and 150 with textured implants; of the BIA-ALCL patients with available data, 12 of 12 (100%) had been exposed to textured implants. All but one patient presented with a radiologically detectable pericapsular effusion (15/16, 93.8%). Definitive pathologic staging information was available for 15 of 16 (93.8%) patients, with 10 patients presenting with a tumor stage of T1, 3 with T2, 1 with T3, and 1 with T4. No patients had any pathologic evidence of regional lymph node involvement. Sample types submitted for FC included fluid and tissue specimens, and the diagnostic procedure types included cytology specimens, including bedside or clinic-acquired fluid specimens (n = 101), OR-acquired fluid specimens (n = 46), and capsulectomies (n = 143).

Table 1.

Clinicopathologic features of study cohort

Clinical features	BIA-ALCL Patients (n = 16)
Age (median years, range)	62 (38–73)
Sex (M/F)	0/16
Side (Left/Right)	10/6
Textured implants (number, %)	12/12 (100%)
Presented with effusion	15/16 (93.8%)
At least 1 positive flow specimen (number, %)	15/16 (93.8%)
At least 1 positive cytology specimen (number, %)	13/13 (100%)
Positive capsulectomy specimen (number, %)	10/16 (62.5%)
Pathologic staging	Known in 15/16
T1	10/15 (66.7%)
T2	3/15 (20%)
T3	1/15 (6.7%)
T4	1/15 (6.7%)
N0	15/15 (100%)
FC sample sources	All FC samples (n = 290)
Fluid specimens
Bedside/clinic sourced (number, %)	101/290 (34.8%)
OR sourced (number, %)	46/290 (15.9%)
Capsulectomy specimens (number, %)	143/290 (49.3%)
Adequate FC samples	All FC samples (n = 290)
Fluid specimens
Bedside/clinic-acquired (number adequate, %)	78/101 (78%)*
OR-acquired (number adequate, %)	35/46 (79.5%)*
Capsulectomy specimens (number adequate, %)	60/143 (42%)
Patients with at least 1 adequate FC sample	119/182 (65.4%)
Immunophenotypic features of BIA-ALCL	Patients with positive FC (n = 15)
Surface CD3 (number, %)	3/15 (20%)
Cytoplasmic CD3 (number, %)	0/4 (0%)
CD2 (number, %)	11/15 (73.3%)
CD5 (number, %)	7/15 (46.7%)
CD7 (number, %)	6/15 (40%)
CD4 (number, %)	15/15 (100%)
CD8 (number, %)	3/15 (20%)
CD30 (number, %)	15/15 (100%)
CD45 (number, %)	14/15 (93.3%)
CD56 (number, %)	6/15 (40%)
CD279/PD-1 (number, %)	1/15 (6.7%)

^*P < 0.0001 compared to capsulectomy specimens (Fisher’s exact test)

The BIA-ALCL samples showed immunophenotype similar to those previously described^19–21,23, with 22/22 (100%) of BIA-ALCL samples expressing CD30, with 3/22 (13.6%) expressing surface CD3, 15/22 (68.2%) expressing CD2, 9/22 (40.9%) expressing CD5, 6/22 (27.2%) expressing CD7, 22/22 (100%) expressing CD4, 3/22 (13.6%) expressing CD8, 21/22 (95.4%) expressing CD45, 9/22 (40.9%) expressing CD56, and 3/22 (13.6%) expressing CD279/PD-1. Expressed in terms of the 15 patients whose BIA-ALCL was accurately detected by FC, 15/15 (100%) expressed CD30, with 3/15 (20%) expressing surface CD3, 11/15 (73.3%) expressing CD2, 7/15 (46.7%) expressing CD5, 6/15 (40%) expressing CD7, 15/15 (100%) expressing CD4, 3/15 (20%) co-expressing CD8 (20%), 14/15 expressing CD45 (93.3%), 6/15 (40%) expressing CD56, and 1/15 (6.7%) expressing CD279/PD-1.

One case (6.7%) showed a so-called “null phenotype,” negative for all surface markers except for CD30. Separate molecular testing for T-cell clonality in this case showed evidence of a clonal TCR gene rearrangement. Four patients were additionally tested with an additional FC assay to evaluate cytoplasmic CD3 because of absence of surface CD3. All were negative (0/4, 0%) but all these patients expressed at least one other T-cell marker, usually CD2. Of the 6 cases that were CD56 positive, one showed bright, uniform expression brighter than background NK-cells²⁴, one case showed moderate expression at the level of background NK-cells, and the remaining 4 cases showed very dim, partial expression, slightly above background intensity but well below the level of background NK-cells.

Fifteen of the 16 patients with BIA-ALCL underwent capsulectomy procedures at our institution as part of standard disease management. Of these patients, 11 patients had morphologic and immunohistochemical findings consistent with BIA-ALCL detected in their capsulectomies. Another 4 patients showed no morphologic or immunohistochemical evidence of involvement by BIA-ALCL in the capsulectomy tissue pathologic evaluation, despite extensive tissue sampling. However, all 4 of these patients had positive cytology with morphologic and immunohistochemical findings consistent with BIA-ALCL in the peri-capsular fluid drained during their capsulectomy procedures. These findings suggested that these 4 patients had low stage disease that was confined to the fluid around the implant, without evidence of tissue involvement.

3.2. Sample viability and inadequate sample rates vary by specimen type and source

Of the overall 290 FC samples assessed, 173 (59.7%) were considered adequate by the criteria outlined previously in the Methods section, and 117 were inadequate. The average total cellularity of the adequate samples was 11.8 million cells; the average viability was 63.0%. Of the 101 FC tests performed on bedside or clinic-acquired liquid specimens, 77.2% were adequate, while 22.8% were inadequate. The average cellularity was 8.71 million cells, and the average viability was 85.4%. In 46 OR-acquired fluid specimens, FC studies for 76.1% were adequate, while 23.9% were inadequate, with average cellularity of 19.6 million cells and 69.3% viability. The 143 capsulectomy tissue samples showed lower adequacy rates; FC studies for 42% were adequate by our criteria compared to 58% inadequate. The average cellularity for capsulectomy samples was 3.37 million cells with 39% viability. The rate of adequate FC results in capsulectomy specimens was significantly decreased compared to both bedside/clinic-acquired cytology (P < 0.0001) and OR-acquired (P < 0.0001) fluid specimens by Fisher’s exact test. There were 3 cases initially reported as “atypical” because of unusual findings, which were reviewed separately from the other 173 adequate cases. Flow cytometric findings for these cases are also summarized in Supplemental Figures 1–3.

Of the cases considered inadequate for FC assessment by our criteria, 34 came from fluid specimens. Of these, 30 of these samples had concurrent cytologic assessment of these fluid specimens, 90% of which were reported as diagnostic specimens. The remaining inadequate FC specimens came from capsulectomy tissue samples, all of which were considered adequate for diagnosis. All the 182 patients in our cohort had either a diagnostic cytologic specimen, capsulectomy specimen, or both.

3.3. Test efficacy of flow cytometry

Efficacy calculations are summarized in Figure 2. Of the total 290 samples, 173 were considered adequate. In 182 patients tested, including 16 with a definitive diagnosis of BIA-ALCL made by morphology, 119 patients (65.4%) had at least one FC test considered diagnostic. Fifteen patients with BIA-ALCL had at least one positive FC result and one had only false-negative FC results. However, of these 15 patients with positive FC results, 5 patients had at least one (ranging from 1–2) false negative result in either prior or subsequent specimens. There were no false positive FC results. There were no BIA-ALCL patients whose diagnosis was initially made on FC alone. In most patients (11/16), a definitive diagnosis of BIA-ALCL was first rendered on a cytology specimen taken in the clinic or at bedside. In 10 of those patients, a diagnosis was made on the first cytology sampling; one patient required a second FNA procedure for a definitive diagnosis. Four patients’ initial diagnoses were made during definitive capsulectomy surgery, either via cytologic evaluation of aspirated seroma fluid, histologic evaluation of capsulectomy specimen, or both. In one patient, initial diagnosis was made via histologic assessment of a core biopsy of a peri-capsular mass.

Figure 2.

Contingency tables and testing performance.

Using the total number of FC specimens to assess test efficacy, the sensitivity was 75.9%, specificity was 100%; PPV was 100%, and NPV was 95.4%. Considering the 119 patients with adequate and diagnostic FC samples, and FC sensitivity for individual patients was 81.3% while specificity was 100%; the PPV was 100% while NPV was 97.2%. Several patients had both true positive and false negative results on different specimens, accounting for this difference in efficacy. Receiver operating characteristic curve based on FC sensitivity and specificity data from the 119 patients tested showed an area under the curve of 0.897 (95% Confidence Interval 0.808 to 0.985, P < 0.0001) for all tests.

3.4. Test efficacy in different sample types

We also assessed test efficacy between different sources of adequate FC specimens (clinic-acquired liquid specimens, OR-acquired liquid specimens, and capsulectomies). In 78 adequate clinic-acquired liquid specimens, we calculated a sensitivity of 77.8%, a specificity of 100%, a PPV of 100%, and a NPV of 97.2%. In 35 adequate OR-acquired liquid specimens, we calculated a sensitivity of 75%, specificity of 100%, PPV of 100%, and NPV of 88.5%. In 105 adequate capsulectomy specimens, we calculated a sensitivity of 75%, specificity of 100%, PPV of 100%, and NPV of 96.3%.

We also assessed if a combined capsulectomy and OR-acquired liquid specimen were more sensitive than a liquid specimen alone. There were 2 patients with false negative FC results in OR-acquired liquid specimens; neither of these patients had subsequent positive FC results on capsulectomy specimens. Conversely, 3 patients with positive FC results on OR-acquired liquid specimens had subsequent false negative FC results on capsulectomy specimens.

3.5. Review of false negative cases

We also reviewed false negative and atypical FC cases to gain insights into diagnostic pitfalls in FC assessment. Three pathologists (AC, MR, and OL) re-reviewed 7 false negative and 3 atypical cases in a blinded manner, randomly mixed with true positive and true negative cases. False negative cases were considered “corrected” if 2 of the 3 author reviewers identified an abnormal CD30-positive population. Upon repeat review, 6 of the 7 false negative cases (85.7%) were correctly diagnosed as positive, suggestive of a diagnostic error in the initial interpretation. In all 173 adequate samples, this improved the sensitivity of the FC assay to 96.6% and the NPV to 99.3%, with specificity and PPV remaining at 100%. For the 119 patients’ overall results, only one patient had entirely false negative results after the authors repeated review, improving sensitivity to 93.8% and NPV to 99%, with specificity and PPV remaining at 100%.

The 7 false negative cases were from 5 patients. Of these, 4 patients’ disease staging showed BIA-ALCL confined to either the capsular fluid or the luminal side of the capsule. Of these 4 patients, 2 of them had disease only detectable in capsular fluid by cytologic review, with no evidence of disease on histopathologic review of the capsule tissue, despite extensive tissue sampling. The 2 patients with histopathologic evidence of disease on the luminal side of the capsule had only focal involvement. The 5th patient, whose false negative cases were not resolved by the consensus review, had BIA-ALCL infiltrating the capsule in aggregates. Of note, this patient’s capsulectomy specimen showed evidence of prominent tumor necrosis, and FC testing of her OR-acquired fluid and capsulectomy tissue samples showed limited viability of 35.7% and 23.7% respectively. We hypothesized that extensive necrosis and debris led to false negative diagnoses in this patient’s FC samples.

A careful review of corrected false negative diagnoses showed that the neoplastic cell populations were frequently missed due to early gates in the FC assessment hierarchy, such as when using scatter parameters to select for singlet events, viable cells, or mononuclear cells. In these cases, neoplastic cell populations were inadvertently excluded from downstream FC analysis or greatly reduced in number available for analysis. For example, on a FSC area versus FSC height plot, which is often used to gate for singlet events, BIA-ALCL cell populations were often located in areas containing increased doublets or very large events, which are frequently gated out of further analysis (Figure 3A). Likewise, on a FSC versus SSC plot, often used to gate viable cells and mononuclear cells, the neoplastic cells in BIA-ALCL were often in the same area as granulocytes, resulting in exclusion from downstream analysis (Figure 3B–C). This also tended to occur in a context of extensive debris and non-specific antigen binding despite attempts to gate these confounding events out using early gates, likely causing further confusion in distinguishing true neoplastic events from nonspecific events. Expansion of singlet, viable cell, and mononuclear cell gates to include more cells allowed for more neoplastic cell events in later gates and improving analyst detection rates (Figure 3D–F).

Figure 3.

Adjusting gating strategies to avoid false negative findings. A. In this example, the initial singlet gate is too small, preventing neoplastic cells/events from appearing in downstream analysis. B. In addition, gating for mononuclear cells is also too restrictive. C. The result of the overly restrictive early gates is that the CD30 bright, surface CD3 negative neoplastic cells are too few to recognize, resulting in a false negative diagnosis. D. The same case now has a corrected initial singlet gate to include larger events. E. Expanded mononuclear cell gate to include events with higher side scatter. F. The CD30 positive population is much more evident when the initial gates are more permissive.

We also separately reviewed the 3 cases reported as “atypical.” One case was in a patient with BIA-ALCL and considered a false negative case (see Supplemental Figure 1); a repeat review identified an abnormal CD30 positive population with dim expression of other pan-T-cell antigens. The population was undetected due to the reasons previously described. The second case reported the presence of CD30-positive T-cells with otherwise normal phenotype, which we considered a true negative case (see Supplemental Figure 2). The third case detected a CD30-negative T-cell population, which was ultimately considered a reactive population, and we considered this case a true negative (see Supplemental Figure 3).

4. DISCUSSION

Flow cytometry is a vital adjunct test in lymphoproliferative disorders. In BIA-ALCL, the mainstay of initial diagnosis is morphologic evaluation, either by cytologic assessment of effusion fluid or biopsy of mass lesions.^6,14 While BIA-ALCL is an uncommon outcome of breast implant exposure and the use of the implicated textured implants is declining, implantation of textured devices continues despite knowledge of BIA-ALCL, and many still carry these devices.^11,12 While our findings show that FC has very high specificity and relatively high sensitivity, FC is never the sole modality by which a patient is diagnosed with BIA-ALCL, and a significant proportion of FC tests for BIA-ALCL do not contain an adequate number of viable cells for a confident assessment. However, there are important benefits of FC, including rapid turnaround time and providing additional evidence of disease, which can be particularly helpful in cases with challenging or equivocal cytology findings. This information can be used to plan for further diagnostic procedures or to allow more precise planning of definitive capsulectomy surgery, as a complete resection with uninvolved surgical margins is essential for management.^14,18,25

Sample source appears to be an important determinant in the diagnostic utility of FC samples. Liquid samples show better viability and adequacy rates compared to capsulectomy specimens. While we could not systematically evaluate the etiology of these discrepancies, we hypothesize that they are related to pre-analytic variables, including time to testing. In our institution, liquid specimens for FC testing are placed into RPMI media during procedures and transported quickly to the FC laboratory. Capsulectomy specimens are submitted fresh and unfixed to a pathology grossing room, requiring additional time to reach appropriate media and the FC laboratory because of specimen processing workflow. Capsulectomy specimens also can be sparsely cellular with only patchy involvement by BIA-ALCL or involvement only in the peri-implant fluid, without evidence of neoplastic cells in capsulectomy sections, even after extensive tissue sampling for microscopy and immunohistochemistry.^6,26,27 Therefore, we consider capsulectomy tissue specimens the least valuable for FC, especially if there is a concurrent OR-acquired fluid specimen to analyze. However, FC of the capsule tissue should be considered in the absence of other fluid specimens, and histologic assessment of the capsule is still vital for staging purposes.¹⁴

Our assessment of false negative FC results provides some insight into possible diagnostic pitfalls. As noted in previous studies, we found that the neoplastic cells in BIA-ALCL tend to show high light scatter characteristics due to cell size and complexity, somewhat similar to the Hodgkin/Reed-Sternberg cells of classic Hodgkin lymphoma.^{20,23,28–31} Many FC analysis techniques rely on the gating of singlet events based on scatter characteristics; as BIA-ALCL cells often fall into areas with high scatter and these cells may be inadvertently gated out of the downstream analysis.²⁹ Therefore, we emphasize the importance of early gates in any FC gating hierarchy. Flow cytometric assessment of T-cells should also not be limited to T-cells with an expression of surface CD3, as this can be a source of additional diagnostic challenges; only a minority of our BIA-ALCL cases were positive for surface CD3, and those we tested were also negative for cytoplasmic CD3 expression.³² Additional markers such as CD25 and CD26 may also be useful in FC assessment of BIA-ALCL.^19,33 Nonetheless, our cohort had one patient with only false negative results after consensus review, which further emphasizes the need for additional cytologic or histopathologic evaluation.

The findings of this study suggest that FC for BIA-ALCL is a useful adjunct or confirmatory diagnostic test that is extremely specific in the right contexts. However, FC is not an ideal first-line test, and its use as the sole diagnostic modality should be avoided because of high rates of inadequate samples, particularly when testing capsular tissue specimens. Flow cytometry also does not affect the gold-standard efficacy of histologic and cytologic evaluation. There are additional limitations of FC including costs, test complexity, and required expertise. Therefore, cytologic or histopathologic diagnosis should remain the gold standard for this entity.⁶ Flow cytometry performed on a capsulectomy specimen is of limited value, as these specimens have reduced viability and a significantly higher rate of inadequate samples compared to FC testing of liquid samples. In limited cases where FC was performed on both liquid samples and capsulectomy specimens acquired via surgery, FC of capsulectomy samples added no diagnostic value. Fluid specimens are the preferred sample type for FC of BIA-ALCL and can provide additional diagnostic value to cytopathology assessment but should not be employed as the only diagnostic modality.