Clinical outcome and Ki67 evaluation in dogs with nodal small cell B-cell lymphoma diagnosed by flow cytometry

2.1 Study sample

Dogs with nodal small cell B-cell lymphoma diagnosed by flow cytometry were identified through the Colorado State University-Clinical Hematopathology Laboratory. Cases had flow cytometry performed on a lymph node aspirate and an expansion (>80% of total cells) of small-intermediate sized CD21+ B-cells with a median B-cell FS <500. Criteria rationale and a figure depicting B-cell FS in a case compared to a healthy dog lymph node and DLBCL case are provided in the Supporting Information S1. As a size comparison, the mean FS of canine neutrophils in the peripheral blood is 620 (SD, 61).⁵ From July 27, 2017 to September 27, 2018, we measured the percentage of Ki67-expressing B-cells by flow cytometry for cases with adequate sample remaining after routine immunophenotyping. Ki67 expression was measured for research purposes and not reported, so these results did not change treatment decisions. Veterinarians submitting samples for all cases with Ki67 expression data were contacted to provide medical records.

2.2 Flow cytometry

Routine immunophenotyping was performed on a fresh lymph node aspirate as previously described, using the antibody panel in Table 1.¹³ Samples were analyzed on a 3 laser Coulter Gallios instrument (Beckman Coulter Inc., Brea, California) and data were analyzed using Kaluza Analysis Software (Beckman Coulter Inc.). Cell size was approximated by median linear FS. Expression of CD21 and class II major histocompatibility complex (MHC) on B-cells was measured by median fluorescence intensity and the percentage of CD25-expressing B-cells was determined as previously described.¹⁴ Control CD4 T-cell size, B-cell size and B-cell CD21, class II MHC and CD25 expression data from lymph nodes of 37 healthy dogs with no evidence of lymphoproliferative disease by a combination of histology, flow cytometry or PCR for antigen receptor rearrangements were used as comparison. CD25 expression data were available for 23/37 healthy control dogs. For all lymphoma cases, intracellular Ki67 (SolA15 clone, eBioscience, San Diego, California) was measured in CD21+ B-cells by flow cytometry on the same lymph node aspirate used for routine immunophenotyping as previously described.¹² We gated on CD21+ B-cells, excluding CD5+ T-cells and CD18+ neutrophils, to measure the percentage of B-cells positive for Ki67 (Ki67%). We previously measured Ki67 expression on control B-cells from healthy dog blood and lymph node samples (n = 12).¹²

2.3 Clinical variables

Signalment, physical exam and imaging findings, and hematologic and biochemistry data were extracted from the Colorado State University-Clinical Hematopathology Laboratory submission form and medical record. Laboratory abnormalities were identified based on the reference intervals for the laboratory performing the CBC or biochemistry panel. Thrombocytopenia was defined as a platelet count below the reference interval without evidence of platelet clumping on blood film review. Staging was performed at the discretion of the supervising clinician and not all dogs were fully staged. Minimum stage was assigned based on the supervising clinician's assessment in the medical record; if no stage was assigned, a minimum stage was assigned by the author JHB based on recorded physical examination findings and review of diagnostics performed. Determination of whether dogs with lymphocytosis or circulating blasts were stage V or not was variable between supervising clinicians and for consistency, all dogs with a lymphocytosis or circulating blasts were assessed as stage V. Substage was determined by the supervising clinician or the author JHB based on recorded clinical signs at the time of diagnosis.

2.4 Cytology and histology

All available lymph node cytology and histology reports that were performed by a board-certified pathologist were reviewed. Cytology slides were not available for re-evaluation. For lymph node cytology reports, the diagnosis, lymphocyte cell size, size determination method, chromatin pattern, nucleoli description and cytoplasm description were recorded by author EDR. For histology reports, the tumor architecture pattern, cytomorphologic features and mitotic rate were recorded by author KLH.

2.5 Statistical analysis

Clinical and flow cytometry data were summarized. For continuous variables, normality was assessed by a Shapiro-Wilk test. Continuous data were expressed as median and range or mean and SD and categorical data were summarized by percentages. Flow cytometry data were compared between lymphoma cases and control B-cells using a Mann-Whitney test. Treatments were categorized based on the induction protocol used. Response to treatment was evaluated by the attending clinician or author JHB using available data in the medical record. Treatment responses were classified as complete response (CR), partial response (PR), stable disease or progressive disease. Veterinary Cooperative Oncology Group (VCOG) nodal response criteria were used when lymph node measurements were available.¹⁵ Progression free survival (PFS) was calculated from the date that treatment was initiated to the date of progressive disease. Overall survival (OS) was calculated from the date of lymphoma diagnosis to the date of death. In cases with a definitive lymphoma diagnosis by cytology, the date of cytologic diagnosis was used. In cases without cytology or an equivocal cytologic diagnosis, the date of flow cytometry diagnosis was used. Cases alive at the time of data collection or lost to follow up were censored. Median PFS and OS were calculated by the Kaplan-Meier method. Clinical variables and flow cytometry variables were tested for association with PFS and OS by univariable analysis using Kaplan-Meier log rank tests. For continuous risk factors, X-Tile Software was used to predict the best cutoffs to divide cases based on outcome and these cutoffs were used for survival analysis.¹⁶ Multivariable Cox proportional hazards regression analysis was performed to investigate variables associated with PFS and OS. Four multivariable models were performed: (a) PFS model with treatment type (CHOP-based protocol including cyclophosphamide, doxorubicin, vincristine and prednisone vs no treatment or other treatment) included as a variable; (b) PFS model with treatment type excluded; (c) OS model with treatment type included; (d) OS model with treatment type excluded. Variables with a P-value < .25 testing association with PFS and OS across all cases by univariable analysis were passed into the PFS and OS multivariable models, respectively. Variables were added into the model backward; first all variables with P < .25 were added and then variables with a P-value > .1 were removed sequentially. Statistical analysis was performed in GraphPad Prism version 9.1.2 (San Diego, California) and MedCalc Software version 20.015 (Ostend, Belgium) and 2-sided P-values < .05 were considered significant.

3.1 Study sample

Ki67 expression was measured in 103 cases of nodal small cell B-cell lymphoma from July 27, 2017 to September 27, 2018 and cases were contacted for follow up information. Medical records were obtained for 54 cases, but 5 cases were omitted from the study because of inadequate follow up or a prior diagnosis of lymphoma and chemotherapy treatment. Specifically, 2 cases had no follow up data, 1 case had a prior diagnosis of lymphoma treated with COP (cyclophosphamide, vincristine, and prednisone) chemotherapy and flow cytometry was performed at relapse, 1 case had a history of splenectomy and subsequent lymphoma treated with doxorubicin and chlorambucil, and 1 case had a history of lymphocytosis and was undergoing chlorambucil and prednisone treatment when generalized lymphadenopathy developed and flow cytometry was pursued. The remaining 49 cases were included in the study.

3.2 Clinical presentation

Information including signalment, stage, substage, hematologic data and biochemical data for dogs enrolled in this study are summarized in Table 2. The median age at diagnosis was 10.3 years (range, 2.1-14.8 years) and median body weight was 19.2 kg (range, 2.7-62.2 kg). Most cases (96%) were assessed as minimum stage III or higher. Eighteen cases (37%) had stage V disease, which was attributed to lymphocytosis or circulating atypical lymphocytes in 16 cases, pulmonary involvement in 1 case, and presence of a mediastinal mass in 1 case. Eleven cases (22%) were assessed as substage B. Among the anemic cases (n = 15), the median hematocrit was 32% (range, 25%-37%). Among the thrombocytopenic cases (n = 13), the median platelet count was 138 000/μL (range, 3000-176 000/μL). Four cases had concurrent anemia and thrombocytopenia. One case had pancytopenia without lymphocytosis. Nine of 10 dogs with neutrophilia had mild neutrophilia while 1 case had 45 600 neutrophils/μL. Among cases with lymphocytosis, 12/14 had a lymphocyte count <35 000/μL and 2 cases had a marked lymphocytosis >100 000/μL.

3.3 Flow cytometry

Flow cytometry data were compared between lymphoma cases and control lymphocytes from lymph nodes of dogs without lymphoproliferative disease (Figure 1). The mean B-cell size in lymphoma cases, as determined by FS, was 407 (range, 314-474), which is 1.1× the mean size of control B-cells from healthy dog lymph nodes (mean FS 362; range, 301-458 SD 36). The percentage of CD25-expressing B-cells was significantly higher in lymphoma cases compared to control lymph nodes from healthy dogs (P < .001). B-cell class II MHC and CD21 expression were not significantly different between cases and controls. Control samples from healthy dog lymph nodes (n = 7) and blood (n = 5) had 7%-26% Ki67-expressing B-cells (median, 8.5%), with 11/12 controls having Ki67% <20%, which was the threshold used to define low Ki67 expression in 2 previous studies.^{11, 12} B-cell lymphoma cases had significantly higher Ki67% than controls (P < .001), though there was a wide range of Ki67% values among lymphoma cases (median, 41%; range, 3%-97%).

3.4 Cytology and histology

Thirty-seven cases had lymph node cytology interpreted by a board-certified pathologist and a cytology report available. Most cases (31/37) had cytology performed within 2 days of flow cytometry (range, 0-21 days). Cytology report data from clinical pathologists at different institutions are summarized in Table 3. Most cases (65%) were diagnosed with lymphoma, 27% of cases were diagnosed as possible or probable lymphoma, and 8% were diagnosed with lymphoid hyperplasia. Among cases diagnosed as possible, probable or definitive lymphoma, 2 cases were interpreted as large cell lymphoma cytologically; these cases had a median B-cell FS by flow cytometry of 365 and 460. The criteria used to categorize cell size were variable. Chromatin patterns were highly variable and included the terms finely stippled, finely granular, finely or slightly clumped, stippled, coarse, fine, and smooth. Nucleoli or nucleolar rings were mentioned in 85% of cases, with most cases having indistinct, faint or variably prominent nucleoli.

Four cases had lymph node biopsy. Tissue was not available for evaluation, but histopathology reports were reviewed. None of the cases were diagnosed as large cell lymphoma, but rather all had small to intermediate-sized cells primarily with condensed chromatin and more rarely with small nucleoli. The histology reports were most suggestive of a variety of small to intermediate-sized mature B-cell lymphoma subtypes including nodular and diffuse forms. None of the cases received a definitive WHO classification but 1 case had differentials of MCL or CLL/SLL and 1 case had differentials of marked follicular hyperplasia or possibly nodular lymphoma and subsequent clonality testing showed clonal immunoglobulin gene rearrangements. For 3 cases with a mitotic rate reported, mitotic activity by microscopy was low to intermediate and the B-cell Ki67% ranged from 6% to 23% by flow cytometry.

3.5 Clinical outcome and survival analysis

The median PFS was 119 days across all 49 cases and 15 cases were censored (Figure 2A). The median OS was 222 days and 8 cases were censored at the time of data collection (Figure 2B). Dogs received a variety of first-line therapies including CHOP-based protocol (n = 32), single agent treatment with either chlorambucil (n = 5), lomustine (n = 2), doxorubicin (n = 2) or rabacfosadine (n = 1) with L-asparaginase and prednisone variably included in these protocols, or prednisone +/− L-asparaginase (n = 6). One dog did not receive any lymphoma specific treatment. Among CHOP-treated cases, 16 dogs (50%) achieved a CR, 10 dogs (31%) achieved a PR, and 5 dogs (16%) had stable disease. Best response to therapy was not available for 1 case. Among CHOP-treated cases, the median PFS was 70 days (5 censored) and the median OS was 267 days. Twenty-two dogs initially treated with a CHOP-based protocol were treated with ≥1 rescue protocol; lomustine +/− L-asparaginase (n = 19) and rabacfosadine (n = 10) were the most pursued rescue therapies. There was no significant difference in PFS (P = .27) or OS (P = .38) between CHOP-treated cases and non-CHOP treated cases (receiving no treatment or other treatment).

Five dogs had concurrent co-morbidities including splenic histiocytic sarcoma, pulmonary carcinoma, suspected thyroid tumor, poorly controlled diabetes mellitus and IRIS grade IV acute kidney injury (relationship to lymphoma diagnosis unclear). The dog with concurrent histiocytic sarcoma was treated with CCNU, the dog with the AKI was treated with L-asparaginase and prednisone, and the remaining 3 dogs were treated with CHOP chemotherapy. The median OS for these 5 dogs was 37 days (range, 5-434 days) which was not significantly different from OS of the other dogs (n = 44) in the study without significant comorbidities (222 days, P = .09).

Potential prognostic factors were evaluated for association with OS by univariable analysis (Table 4). Each variable was evaluated across all cases (n = 49) and cases treated with a CHOP-based protocol (n = 32). For continuous variables, the best predicted cutoff value using X-Tile Software was used to divide groups (Table 4).¹⁶ B-cell Ki67% was prognostic among CHOP-treated dogs, with very low Ki67 expression cases (≤11% Ki67-expressing B-cells) having significantly longer OS (MST 542 days) than cases with Ki67% >11% (MST 242 days; P = .01; Figure 3A). However, the number of cases with Ki67% ≤11% was small (n = 5). When Ki67% was assessed across all cases of all treatment types, there was a similar difference in MST (542 days vs 176 days) but the association with OS was no longer statistically significant (P = .08). Smaller B-cell size as determined by flow cytometry was associated with shorter survival among CHOP-treated cases, but not across all cases. B-cell class II MHC expression and CD25 expression were associated with survival among all cases, but not CHOP-treated cases only. Across all cases, younger dogs (≤7 years) had significantly longer survival than older dogs. This association was not significant among CHOP-treated cases. There was no significant difference in age between CHOP-treated and non-CHOP-treated cases or between cases that did or did not receive a rescue protocol. Among all cases (Figure 3B) and CHOP-treated cases, dogs assessed as substage B had significantly shorter OS than dogs in substage A. Among CHOP-treated cases, the presence of lymphocytosis was associated with significantly shorter OS. Among CHOP-treated cases, dogs that achieved a CR had significantly longer PFS and OS than dogs that achieved a PR or had stable disease. Among cases with lymphocyte cell size assessed on lymph node cytology (n = 32), there was no difference in OS between cases with lymphocytes described as small to intermediate or intermediate-sized and those with intermediate to large or large-sized lymphocytes (P = .54).

Multivariable analysis results are provided in Table 5. Though treatment type (CHOP-based protocol vs other or no treatment) did not significantly affect OS by univariable analysis (P = .38), the multivariable analysis was performed with and without treatment type included as a factor. Five variables had a P-value < .25 by univariable analysis for PFS and were included in the PFS multivariable model: B-cell Ki67%, B-cell size, B-cell CD21 expression, sex, and substage. Only substage had a P-value < .1 in the PFS multivariable model (P = .02). Six variables had a P-value < .25 by univariable analysis for OS and were included in the OS multivariable model: B-cell Ki67%, class II MHC expression, CD21 expression, CD25%, age and substage. Though hyperglobulinemia had a P-value < .25 by univariable analysis, this variable was excluded because data were only available for a subset of cases. All variables tested in the OS multivariable model, including treatment type, had a P-value < .1 except B-cell Ki67% (P = .33), which was excluded from the final model. Older age, substage B, low B-cell CD21 expression and high B-cell CD25 expression were independently associated with increased likelihood of death irrespective of whether treatment type was included in the model. Low B-cell class II MHC expression was associated with increased risk of death when treatment type was included in the model.