DNA methylation analysis might identify prognostic CpG sites in CHOP-treated dogs with multicentric high-grade B-cell lymphoma (MHGL) with heterogenous prognosis.

Objective

To identify prognostic CpG sites of MHGL through genome-wide DNA methylation analysis with pyrosequencing validation.

Animals

Test group: 24 dogs. Validation group: 100 dogs. All client-owned dogs were diagnosed with MHGL and treated with CHOP chemotherapy.

Methods

Cohort study. DNA was extracted from lymph node samples obtained via FNA. Genome-wide DNA methylation analysis using Digital Restriction Enzyme Analysis of Methylation (DREAM) was performed on the test group to identify differentially methylated CpG sites (DMCs). Bisulfite pyrosequencing was used to measure methylation status of candidate DMCs in the validation group. Median survival times (MST) were analyzed using Kaplan-Meier (log-rank) product limit method.

Results

DREAM analyzed 101 576 CpG sites. Hierarchical clustering of 16 262 CpG sites in test group identified group with better prognosis (MST = 55-477 days vs 10-301 days, P = .007). Volcano plot identified 1371 differentially methylated CpG sites (DMCs). DMC near the genes of FAM213A (DMC-F) and PHLPP1 (DMC-P) were selected as candidates. Bisulfite-pyrosequencing performed on validation group showed group with methylation level of DMC-F < 40% had favorable prognosis (MST = 11-1072 days vs 8-1792 days, P = .01), whereas group with the methylation level combination of DMC-F < 40% plus DMC-P < 10% had excellent prognosis (MST = 18-1072 days vs 8-1792 days, P = .009).

Conclusion and Clinical Importance

Methylation status of prognostic CpG sites delineate canine MGHL cases with longer MST, providing owners with information on expectations of potential improved treatment outcomes.

Abbreviations

CHOP: cyclophosphamide, doxorubicin hydrochloride, vincristine sulfate, prednisone
CpG: 5′-C-phosphate-G-3
DLBCL: diffuse large B-cell lymphoma
DMC: differentially methylated CpG site
DREAM: Digital Restriction Enzyme Analysis of Methylation
MHGL: malignant high-grade lymphoma
PBMC: peripheral blood mononuclear cells

1 INTRODUCTION

Multicentric high-grade lymphoma (MHGL) is the most common type of lymphoma in dogs comprising more than 80%^1-4 of the lymphoma cases,⁵ where 70% of MHGL is B-cell originated. Chemotherapy with CHOP based therapy significantly improves survival time (ST) in MHGL, but relapses are still common.⁶ Several studies have described the factors which delineate prognoses within the subgroup of MHGL, including immunophenotyping,⁷ WHO clinical substages,⁵ anemic status,⁸ and the presence of disease relapse⁹ which are relatively feasible to assess in the clinics. Advanced classification which includes histopathological findings and molecular tests such as immunohistochemistry for argyrophilic nucleolar organization region¹⁰ or survivin,¹¹ expression levels of MHC class II¹² and minimal residual disease¹³ have also been studied to better understand the factors contributing to the heterogeneity seen in MHGL cases. However, these studies have various limitations which confine the practicality in identifying MHGL cases with better prognosis, especially the advanced classifications which requires invasive biopsies. Conversely, clinical data offers expedited and simplified prognosis assessment from routine minimum database. Yet, perception toward the criteria of clinical sub-stages can be subjective and varied between clinicians,¹⁴ which might be contributed mostly by the lack of specific criteria to determine clinical substages.^15-17 Hence, there is a need for a prognostic indicator which is both simple in sample procurement and objective in nature.

Epigenetics is the study of changes in gene expression and function which are heritable yet not associated to any DNA sequence¹⁸ and its application possesses important roles in describing human hematopoietic malignancies including DLBCL.¹⁹ Aberrant methylation of the cytosine in DNA is a promising prognostic tool for human DLBCL, like P16²⁰ and KLF4²¹ because of their silencing consequences resulting from DNA hypermethylation. In human DLBCL, epigenetic heterogeneity in DNA methylation pattern correlates with a poorer prognosis,²² signifying the feasibility of DNA methylation studies in detecting candidate CpG site prognostic biomarkers in dogs with MHGL. In dogs, while there are multiple studies which have utilized next-generation sequencing based methylation analysis to relate molecular signature of DNA methylation and transcriptome with the prognosis of MHGL,^23-25 there is currently only 1 study which described the use of a singular CpG site methylation in the gene DAPK as a possible prognostic marker of nodal high-grade B-cell lymphoma.²⁶ Despite the presence in statistical significance in predicting a poorer outcome of dogs with high-grade-B-cell lymphoma, the study focused only a singular CpG site with a rather smaller study group. Hence, there is still a need for a genome-wide screening of the DNA methylome in canine MHGL to detect candidate CpG sites as prognostic markers which could be superior in delineating cases with a better prognosis.

This study focuses on utilizing genome-wide DNA methylation data to stratify canine MHGL subgroups according to their STs and determine specific differentially methylated CpG sites (DMCs) as candidate prognostic biomarkers of canine MHGL.

2 MATERIALS AND METHODS

2.1 Enrollment

In the test cohort, 24 client-owned dogs with naturally occurring dogs with multicentric high grade B-cell lymphoma cases (MHGL), including centroblastic types according to the updated Kiel classification at the Veterinary Medical Center of the University of Tokyo (UT) between August 2007 and December 2010 were enrolled. Cytology was procured using fine needle aspiration (FNA) biopsy of clinically enlarged lymph nodes. The cases were diagnosed based on clinical finding and were classified according to the cytological findings (high mitotic rate and intermediate to large-sized cells) and staged with reference to the WHO Clinical Staging System for Lymphoma in Domestic Animals.²⁷ The test cohort cases were recruited for the genome-wide DNA methylation analysis using canine Digital Restriction Enzyme Analysis of Methylation (DREAM). The data of DNA methylation patterns of PBMCs from 3 healthy beagle dogs found in a previous study were also included in the genome-wide DNA methylation analysis.²⁸

In the validation cohort, 100 client-owned dogs with naturally occurring canine MHGL cases at private hospitals all over in Japan which cytology slides has been sent to a diagnostic company (North Lab, Sapporo, Japan) were enrolled. Cytology was procured using fine needle aspiration (FNA) biopsy of clinically enlarged lymph nodes. The cases were diagnosed as MHGL based on clinical finding and were classified according to the cytological findings (high mitotic rate and intermediate to large-sized cells) and staged with reference to the WHO Clinical Staging System for Lymphoma in Domestic Animals²⁷ by a board-certified veterinary pathologist (T.I). The validation cohort cases were enrolled for bisulfite conversion and pyrosequencing (bisulfite-pyrosequencing) to detect the DNA methylation level of the CpG sites of target DMCs identified through canine DREAM analysis. PCR for Antigen Receptor Rearrangements (PARR) using previously described methods,^{29, 30} was performed to identify the immunophenotype of each case.

Other inclusion criteria for each study were as follows: (a) primary nodal lymphoma, (b) the use of CHOP (vincristine, cyclophosphamide, doxorubicin, and prednisolone)-based chemotherapy for treatment, and (c) the availability of preserved genomic DNA derived from lymphoma cells at diagnosis before treatment or cytology samples used for diagnosis. Clinical staging was determined in this study as follows: stage I, involvement of only a single node; stage II, involvement of multiple lymph nodes in a region; stage III, generalized lymph node involvement; stage IV, liver and/or spleen involvement (suggested by abdominal ultrasound and/or cytology); stage V, presence of circulating lymphoblasts or immature lymphocytes in the peripheral blood on the blood smears. Substage b was defined as the presence of any of the following: (a) moderate to severe lethargy, (b) moderate to severe decreased appetite, (c) vomiting or diarrhea, and (d) dyspnoea associated with lymphoma. Survival times were determined from the reports of the primary veterinarian and review of the medical records of MHGL cases.

A signed informed consent was obtained from the owners of each dog for both cohorts adhering to ethical guidelines. Cytology slides for DNA extraction of the validation cohort were only used at least 3 years after the diagnosis, and the usage of the specimens was approved by the board of directors of the commercial pathology laboratory (North Lab, Sapporo, Japan.) Genomic DNA was extracted from 24 and 100 cytology slides of lymphoma cases for canine DREAM analysis in test cohort and bisulfite-pyrosequencing in validation cohort respectively using the DNeasy Blood & Tissue Kit (Qiagen, Hilden, Germany).

Details of all the cases above are available in the Table S1.

2.2 Digital restriction enzyme analysis of methylation (DREAM)

Genome-wide DNA methylation analysis using next-generation sequencing was performed as described in previous studies.^{28, 31-33} Genomic DNA (1 μg) extracted from the samples was mixed with 2 pg of a set of artificial standards for calibrators (methylation levels of 0%, 25%, 50%, 75%, and 100%). These mixes were digested with 100 U SmaI endonuclease (New England Biolabs) for 3 h at 25°C, then 50 U XmaI endonuclease (New England Biolabs) were added, and digestion continued for an additional 16 h at 37°C. The digested DNA was purified using Agencourt AMPure XP magnetic beads (Beckman Coulter). The 3′ recessed ends of the DNA created by XmaI digestion were filled in with a dCTP, dGTP, and dATP mix (0.4 mM of each) and 3′ dA tails were added to all restriction fragments by Klenow DNA polymerase lacking 3′-5′ exonuclease activity (New England Biolabs).

Illumina paired-end sequencing adaptors were ligated using T4 DNA ligase (New England Biolabs). The ligation mix was size selected by Dual-SPRI size selection with Agencourt AMPure XP to obtain DNA fragments ranging from 250 to 450 base pairs (bp). Purified DNA was amplified with Illumina paired-end PCR primers using KAPA Hifi Hotstart ReadyMix (Kapa Biosystems) and 11 cycles of amplification. The resulting sequencing library was cleaned with Agencourt AMPure XP beads and sequenced on an Illumina HiSeq 2000 (Illumina).

Sequencing reads were mapped to SmaI/XmaI sites in the canine genome (canFam3.1), and signatures corresponding to the methylated and unmethylated CpGs were enumerated for each SmaI/XmaI site. Methylation frequencies for individual SmaI/XmaI sites were then calculated. The methylation ratio is the ratio of the number of tags starting with CCGGG divided by the total number of rags mapped to a given SmaI/XmaI site.

Methylation levels measured by DREAM were corrected based on values obtained from spikes in standards. Log ratios ln (m/u) and ln (sm/u) were calculated for each standard, where m/u is the expected ratio of methylated and unmethylated reads, whereas sm and u are observed numbers of methylated and unmethylated reads, respectively. Differences in the “expected” minus “observed” log ratios were calculated for each standard. Correction factor c was calculated as an antilogy of the average log difference (expected − observed). Corrected methylation values were then computed as 100% × [(c × sm)/(c × sm + u)] for each CpG site.

At least 20 sequencing reads were used to analyze methylation levels at individual SmaI/XmaI sites. On the basis of technical replicate experiments DREAM can distinguish differences in methylation of >10%, with a false discovery rate (FDR) of 2.4%. We used the definition of CpG islands determined by the University of California, Santa Cruz (UCSC): GC content of ≥50%, length >200 bp, ratio >0.6 of observed number of CG dinucleotides to the expected number on the basis of the number of Gs and Cs in the segment. Sites at the promoter regions are defined as being located within 1 kb from transcription start sites of given genes.

2.3 Identification of target differentially methylated CpG sites (DMCs)

First, we have included CpG sites which were retrievable in all 24 MHGL cases and all PMBCs of healthy beagles to plot meaningful DNA methylation patterns. DNA methylation patterns of the identified CpG sites in the CpG islands (CGI) and non-CGI (NCGI) from canine DREAM analysis were then used to cluster the 24 MHGL cases and 3 healthy beagle dogs into meaningful groups. When good and standard prognosis groups were identified using hierarchical clustering, volcano plot was made using CpG sites which were retrievable in at least 20 of the 24 MHGL cases, with the criteria of absolute difference in methylation ≥20% and P-value <.05 to identify DMCs between good and standard prognosis group. Target DMCs which were significantly different between the good and standard prognosis groups were then selected for further analysis in bisulfite-pyrosequencing.

2.4 Bisulfite-pyrosequencing

Bisulfite-pyrosequencing was used to quantitatively assess DNA methylation for the CpG sites near the genes at Chr 4:29559893 (FAM213A) and Chr1:14232692 (PHLPP1). Briefly, genomic DNA (100 ng) was treated for bisulfite conversion using EZ DNA Methylation-Lightning Kit (Zymo Research, Irvine, California) as per manufacturer recommendation. Bisulfite-treated DNA was then amplified with gene-specific primers which covered the differentially methylated CpG sites identified in Canine DREAM analysis in a 2-step polymerase chain reaction (PCR). Polymerase chain reaction primer sequences and thermal cycling conditions are listed in Table 1. DNA methylation levels were measured as the percentage of bisulfite-resistant cytosine at the CpG sites through pyrosequencing using PSQ24 system with Pyro-Gold reagent Kit (QIAGEN, Hilden, Germany), and the results were analyzed using PyroMark Q24 software (QIAGEN, Hilden, Germany) OR PyroMark Q48 Autoprep (QIAGEN, Hilden, Germany) and analyzed using the provided software PyroMark Q48 Autoprep Instrument Software Version 4.3.3 (QIAGEN, Hilden, Germany).

TABLE 1. PCR primers and thermal cycling conditions.

First step PCR
Target CpG site	Chromosome	Position	Forward primer (5′ to 3′)	Reverse primer (3′ to 5′)
FAM213A	chr4	29 559 894	GGTTGGAAGGAGAGTAGTAGTAAT	ACCCAACAACCACAACACATA
PHLPP1	chr1	14 232 693	AGTAGGTAGTAGGTTTAGGGTAT	ATAACTACCTCTTCCAACTAAAATTTAAA

Second step PCR
Target CpG site	Chromosome	Position	Forward primer (5′ to 3′)	Reverse primer (3′ to 5′)
FAM213A	chr4	29 559 894	GGTTGGAAGGAGAGTAGTAGTAAT	GGGACACCGCTGATCGTTTAACCCAACAACCACAACACATA
PHLPP1	chr1	14 232 693	AGTAGGTAGTAGGTTTAGGGTAT	GGGACACCGCTGATCGTTTAATAACTACCTCTTCCAACTAAAATTTAAA

Pyrosequencing
Target CpG site	Chromosome	Position	Sequence analyzed
FAM213A	chr4	29 559 894	TYGGGATTTT AGATATTYGGATYGGAA
PHLPP1	chr1	14 232 693	YGTTTTYGGG GTTTYGGGGGTTTTT

Note: First step PCR: Thermal cycling conditions: 1 cycle of 94°C for 3 minutes, 38 cycles of denaturation at 94°C for 30 seconds, annealing at 60°C for 40 seconds and extension at 72°C for 40 seconds, final extension of 72°C for 7 minutes and 10°C for storage. Second step PCR: Thermal Cycling Conditions: 1 cycle of 94°C for 3 minutes, 40 cycles of denaturation at 94°C for 30 seconds, annealing at 62°C for 40 seconds and extension at 72°C for 40 seconds, final extension of 72°C for 7 minutes and 10°C for storage.
Abbreviation: PCR, polymerase chain reaction.

2.5 Statistical analysis

In genome-wide DNA methylation analysis, DNA methylation levels between the different groups were compared using the student's t-test. Adjustment for multiple testing in Canine DREAM was performed with the procedure of Benjamini and Hochberg (1995) to identify differential methylation at FDR of 10%. In the test cohort, the groups with different DNA methylation patterns of CpG sites were tested against the median ST of canine lymphoma cases. In the validation cohort, the groups with different DNA methylation levels of target DMCs were tested against the median ST of canine lymphoma cases. Deaths unrelated to lymphoma, loss to follow-up, or cases which were still alive at the point of data analysis were censored. Survival curves were plotted using Kaplan-Meier analysis and tested with log-rank test. Statistical analysis was performed using JMP, Version 16.0.0, SAS Institute Inc., Cary, North Carolina, 1989-2023.

3 RESULTS

3.1 Genome-wide DNA methylation analysis

We deployed DREAM using next-generation sequencing in dogs to identify genomic locations of differentially methylated CpG sites in MHGL with good prognosis. Among 24 samples which were used for the Canine DREAM, 7 to 15 million unique usable reads after conservative filtering (quality filtered and aligned to the dog genome) were successfully generated for DNA methylation analyses. As we used CpG sites consisting of more than 20 reads to assure quantitative ability, the analysis of approximal up to 103 000-145 000 CpG sites of DNA methylation data was possible for each dog.

With the CpG selection criteria mentioned, 16 262 CpG sites in the CGIs and 23 536 sites in the non-CGI (NCGIs) were analyzed. Hierarchical clustering analysis was then performed based on the DNA methylation data of the test cohort and 3 healthy beagles using the CpG sites in the CGI and NCGI separately (Figure 1A; Figure S1). While the analysis of NCGI showed no significance in the clustering of DNA methylation pattern, intriguingly the hierarchical clustering analysis of the CpG sites in the CGI has revealed that 24 cases of MHGL were grouped into 4 different groups (I to IV) with significantly different STs (P = .003) when plotted against the Kaplan-Meier curves (Figure 1A,C). However, 1 of the MHGL cases (circled in red dashed line in Figure 1A) was grouped into the groups with PBMCs of healthy beagles, which was also seen in the NCGI clustering (Figure S1B) and was the same MHGL case. As we further plot the group with the longest ST (good prognosis group, n = 9) against the combined STs of 3 other groups (standard prognosis group, n = 14), stronger significance in difference of ST (P = .007) was shown (Figure 1D). Hence, we were able to extract the group with the longest ST using the DNA methylation pattern of MHGL cases in the CGI.

Details are in the caption following the image — **FIGURE 1**
Open in figure viewer PowerPoint

(A) Hierarchical clustering of the methylation patterns of 16 262 CpG sites in the CpG islands of test cohort. Colors of highlighted groups (green, blue, red, and orange) correspond to the roman numbers listed on the right side of the graph. PBMC data from healthy beagles are not highlighted. One MHGL case was clustered into the healthy beagle group and is circled in red dashed line. (B) Kaplan-Meier Survival Function of 4 groups clustered hierarchically according to CpG site methylation pattern of the CpG-islands with the data of 24 dogs diagnosed with MHGL. Groups and their line colors corresponds to the hierarchical clusters in Figure 1A. (C) Kaplan-Meier Survival Function comparing survival of significantly good prognosis group (Group II in blue) against standard prognosis group (comprising of Group I, III, and IV in red) according to CpG site methylation pattern of the CpG-islands with the data of 24 dogs diagnosed with MHGL.

In order to identify the specific CpG sites which were involved in the DNA methylation pattern associated to MHGL group with good prognosis, we then compared the DNA methylation of all the 101 576 CpG sites (both CGI and NCGI), which has resulted in the detection of 1373 DMCs, where we found 875 differentially hypermethylated CpG sites and 498 differentially hypomethylated CpG sites in the good prognosis group as compared to standard prognosis group as shown in the volcano plot (Figure 2), and the list of top 20 significant DMCs are listed in Table 2.

TABLE 2. The top 20 most significant differentially methylated CpG sites (DMCs) with highest difference in methylation detected in the test cohort.

Nearest gene	Chromosome	Position	Difference in methylation (%)	P-value
AM989461	chr33	21 901 890	−51.9841	.000000137
AB012223	chr25	2 926 298	−46.7576	.0000105
FAM213A	chr4	29 559 894	−46.0331	.0000589
PHLPP1	chr1	14 232 693	−44.8313	.0000926
KF381333	chr9	24 905 237	−44.51233333	.000178445
AB012223	chr25	2 926 083	−40.3403	.00006
DOCK4	chr14	51 695 166	−40.1251	.0000216
ENSCAFG00000023995.2	chr32	3 069 256	−38.34951111	.000201784
GJB6	chr25	17 912 013	−38.2354	.0000479
PINX1	chr25	26 913 178	−36.0974	.000280794
ENSCAFG00000045626.1	chr17	1 661 076	28.229	.000101
ENSCAFG00000042917.1	chr24	44 189 188	34.21984	.000106
ENSCAFG00000048638.1	chr24	45 052 288	35.48118	.0000685
ENSCAFG00000019108.4	chr4	87 939 044	37.69909	.0000129
ENSCAFG00000043834.1	chr20	48 616 896	38.27006667	.000188182
ENSCAFG00000043834.1	chr20	48 617 581	38.32682	.000106
ENSCAFG00000041924.1	chr15	7 195 720	41.19222	.0000102
ENSCAFG00000007606.6	chr1	121 474 361	41.47045238	.000275109
ENSCAFG00000043834.1	chr20	48 617 282	43.749	.000250034
ENSCAFG00000002019.5	chr13	45 559 050	44.42179	.0000775

Note: Candidate DMCs (FAM213A and PHLPP1) for the validation cohort are bolded.

3.2 Bisulfite-pyrosequencing and its relationship with median ST

To validate the data of the DREAM analysis, we selected specific candidate CpG sites from the 1372 DMCs which allows the focal interrogation of DNA methylation on the test groups, namely the CpG site near FAM213A gene at the position of chr4:29559893 (DMC-F) and CpG site near PHLPP1 gene at the position chr1:14232692 (DMC-P) as shown in Figure 2 (colored in blue), using bisulfite-pyrosequencing in our validation group. Among the 100 cases recruited, 13 cases were excluded from the study as they were detected as either T-cell origin or clonally negative in PARR^{29, 30} whereas 1 case was excluded as the ST was not available. DNA methylation of DMC-F in 81 of the 86 samples (93%) was successfully measured, whereas the DNA methylation DMC-P in 80 of the 86 samples (92%) was successfully measured. Only 78 of the 86 samples (91%) had both DNA methylation statuses of DMC-F and DMC-P available. As shown in Figure 3, the mean ± SD of the DNA methylation rate of DMC-F was 51.65% ± 31.25% (95% confidence interval: 44.78%-58.51%) whereas the mean ± SD of the DNA methylation rate of DMC-P was 28.72% ± 32.52% (95% confidence interval: 21.52%-35.91%).

To investigate the association between the percentage of DNA methylation of each CpG site and the ST of the 86 cases, survival curves were plotted using the Kaplan-Meier method and using the cut-off of <40% DNA methylation in DMC-F, <10% DNA methylation in DMC-P as well as the combined DNA methylation <40% in DMC-F and DNA methylation <10% in DMC-P (DMC-F < 4% + DMC-P < 10%). The survival curve comparing the ST between MHGL group with DNA methylation <40% and >40% of the DMC-F showed that the <40% DMC-F group has a significantly longer median survival time (MST) at 499 days (n = 28) as compared to 299 days (n = 53) of the >40% DMC-F group (Figure 4, P = .01). On the other hand, no significant difference in MST was detected between MHGL groups with DNA methylation <10% (MST = 500 days, n = 38) and DNA methylation >10% (MST = 413 days, n = 42) in DMC-P (Figure 5). However, the comparison of ST between MHGL cases with DNA methylation <40% in DMC-F plus DNA methylation <10% in DMC-P and the cases with the remaining DNA methylation patterns (DMC-F < 40% + DMC-P > 10%, DMC-F > 40% + DMC-P < 10%, DMC-F > 40% + DMC-P > 10%), showed that the DMC-F < 40% + DMC-P < 10% group had a greater significance and a longer MST at 692 days (n = 16) as compared to 277 days (n = 62) of the group with combined STs of remaining DNA methylation patterns (Figure 6, P = .009).

Univariate analysis showed that sex, age range, stage and substage were not significantly associated with ST. Only DNA methylation status was significantly associated with ST (P = .004). On multivariate analysis, both DNA methylation status (P < .0001) and the stage of lymphoma (P = .015) were significantly associated with ST in MHGL cases (Table 3). When all the reported prognostic factors were compared between prognosis groups using the Chi-square method, there were no significant difference found between the good and standard prognostic groups (Figure S1A-D).

TABLE 3. Univariate and multivariate analysis of factors affecting survival.

Variable	No. (%) of dog	MST (d)	P-value (univariate)	P-value (multivariate)
Prognosis (n = 86)
Good	16	694 (18-1072)	.001	<.0001
Standard	62	277 (8-1792)
Missing	8
Sex (n = 86)			.85	.12
Male	55	241 (10-1792)
Female	26	356 (11-1228)
Missing	5
Age (n = 86)			.72	.09
<5		166 (1-1792)
5-8		361 (24-1228)
9-12		245 (15-1072)
>13		156 (8-828)
Missing	2
Stage (n = 86)			.34	.02
I	13	359 (89-1792)
II	30	332 (27-1228)
III	19	163 (10-730)
IV	17	244 (1-669)
V	1	406
Missing	6
Sub-stage (n = 86)			.87	.74
a	25	277 (27-993)
b	23	228 (8-1228)
Missing	38

Note: Bold values are the variables which are significantly associated to survival times.

4 DISCUSSION

This study showed that whole-genome DNA methylation analysis could objectively detect prognostic CpG sites which identifies the dogs with MHGL which have significant longer MST when treated with CHOP protocol. The validation cohort further confirmed that selected CpG sites can predict prognosis of dogs with MHGL with bisulfite pyrosequencing of 2 specific CpG sites. Only DNA methylation status was associated to ST on univariate analysis, but DNA methylation status and stage of lymphoma were associated to ST on multivariate analysis.

In the current study, we have utilized the DREAM as a genome-wide screener to detect the DNA methylation changes in dogs with MHGL with the attempt to unravel a novel prognostic biomarker for the disease. While epigenomic studies on the prognostic subtypes of canine diffuse large B-cell lymphoma (DLBCL) are available,²³ the study which used a DNA CpG microarray in canine DLBCL was only able to identify groups associated with long-term survival through cluster of methylation profile, without picking up specific CpG sites which might be prognostic markers. Unlike the previous study,²³ our DNA methylation findings have identified 2 specific CpG sites where DNA methylation profiles have successfully delineated MHGL cases into the good (significantly longer ST) or standard prognosis group. Since prognostic tools for MHGL are still lacking because of the low reliability of clinic-pathological data in assessing treatment responses,^{34, 35} methylation statuses of the 2 CpG sites in this study can be invaluable to clinicians as part of informed consent in decision making by providing evidence of predicted outcomes with CHOP treatments, especially for cases which were stratified into the good prognosis group where clients are reluctant to pursue chemotherapy. Coupled with novel MHGL treatment strategies like chemoimmunotherapy, virotherapy, and guanine nucleotide analogue,^36-38 the lifespan of MHGL cases might be further extended from current life expectancies. On the other hand, as genome-wide methylation studies generally are costly and requires long time for results analysis, DNA methylation profiling of the 2 CpG sites using bisulfite-pyrosequencing in our study gives the advantage of lower cost and shorter turnover time for results in predicting the prognosis for MHGL. Furthermore, our study has exploited the usefulness of DNA which is much stable and easier to handle as compared to RNA samples in the microarray method,²³ making this pyrosequencing technology more practical to the clinical settings as only the cytology sample submitted for clinical pathology assessment is required for testing. Finally, pyrosequencing offers an objective evaluation of the DNA methylation statuses of each sample as the assessment of methylation is fully computerized and can be reproducible as compared to subjective evaluation of clinical pathology by practitioners or specialists.

The interpretation of the pyrosequencing results might require caution by the users. In this study, we set the cut off values of the methylation statuses of DMC-F at <40% and DMC-P at 10% because of the relevance of these points to delineate prognosis groups. It is possible that these cut off values might require adjustments in different populations of dog to effectively delineate prognosis groups, since the epigenome of DLBCL can be heterogenous as seen in human,²² and the distribution of DNA methylation status in both CpG sites is varied in our data (Figure 3). In the current study, we have only selected 2 DMCs to be sequenced as candidate biomarkers in assessing the prognostic outcome of the MHGL cases. Increasing the number of DMCs (as listed in Table 2) to be sequenced will eventually increase the prognostic power of pyrosequencing to predict the outcomes of these cases, but this comes with a downside of increased costs in terms of reagent and labor, as well as the amount of valuable DNA samples. Hence, with consideration of cost-benefit we have determined that the 2 DMCs mentioned above would be suffice in predicting the prognosis of MHGL cases, as confirmed in the survival analysis in Figure 6.

5 CONCLUSION

To conclude, this genome-wide DNA methylation study has successfully detected DMCs which predict the treatment response of dogs with MHGL to CHOP therapy in terms of ST, making them potential prognostic biomarkers which have also been validated through pyrosequencing. These results showed that pyrosequencing is a feasible test to be performed in the clinic which provide a rapid and objective evaluation of the prognostic outcome in dogs with MHGL.

ACKNOWLEDGMENT

This research was supported by the Japan Society for the Promotion of Science (JSPS), Grant-in-Aid for Scientific Research (KAKENHI) Grant Number 20K0623010 and 21H0235103. The study was also funded by Anicom Specialty Medical Institute Inc, Japan. The authors acknowledge the clients and their dogs which have contributed information and samples to this study.

CONFLICT OF INTEREST DECLARATION

A patent application has been submitted (PCT/JP2022/045551).

OFF-LABEL ANTIMICROBIAL DECLARATION

Authors declare no off-label use of antimicrobials.

INSTITUTIONAL ANIMAL CARE AND USE COMMITTEE (IACUC) OR OTHER APPROVAL DECLARATION

Usage of cytology slides was approved by the board of directors of North Lab.

HUMAN ETHICS APPROVAL DECLARATION

Authors declare human ethics approval was not needed for this study.

Supporting Information

Filename	Description
jvim16931-sup-0001-SupplementaryFigure1.pngPNG image, 946.9 KB	Figure S1: Hierarchical clustering of the methylation patterns of 23 536 CpG sites in the non-CpG islands (NCGI) of test cohort. PBMC of beagles are highlighted in orange. One MHGL case grouped into the cluster of healthy beagle PBMC is circled in red dashed line.
jvim16931-sup-0002-SupplementaryFigure2A-B.pngPNG image, 189.1 KB	Figure S2: (A, B) Hierarchical clustering of the methylation patterns of 23 536 CpG sites in the non-CpG islands (NCGI) of test cohort. PBMC of beagles are highlighted in orange. One MHGL case grouped into the cluster of healthy beagle PBMC is circled in red dashed line.
jvim16931-sup-0003-SupplementaryFigure2C-D.pngPNG image, 195.7 KB	Figure S2: (C, D) Chi-square comparison of prognosis groups against sex (A), sub-stage (B), stage (C), and age range (D).
jvim16931-sup-0004-Supplementary_Table_1.pdfPDF document, 96.2 KB	Table S1: Details of cases recruited in each cohort.

Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.

REFERENCES

1Vezzali E, Parodi AL, Marcato PS, Bettini G. Histopathologic classification of 171 cases of canine and feline non-Hodgkin lymphoma according to the WHO. Vet Comp Oncol. 2010; 8(1): 38-49.
10.1111/j.1476-5829.2009.00201.x
CAS PubMed Web of Science® Google Scholar
2Jacobs RM, Messick JB, Valli VE. Tumors of the Hemolymphatic System. Tumors in Domestic Animals. Wiley; 2008: 119-198.
Google Scholar
3Ponce F, Marchal T, Magnol JP, et al. A morphological study of 608 cases of canine malignant lymphoma in France with a focus on comparative similarities between canine and human lymphoma morphology. Vet Pathol. 2010; 47(3): 414-433.
10.1177/0300985810363902
CAS PubMed Web of Science® Google Scholar
4Teske E, Teskel E. Canine malignant lymphoma: A review and comparison with human non-hodgkin's lymphoma. Vet Q. 2011; 16(4): 209-219.
10.1080/01652176.1994.9694451
Google Scholar
5Vail DM, Pinkerton M, Young KM. Hematopoietic Tumors Section A: canine lymphoma and lymphocytic leukemias.
Google Scholar
6Sanz L, López-guillermo A, Martínez C, et al. Risk of relapse and clinico-pathological features in 103 patients with diffuse large-cell lymphoma in complete response after first-line treatment. Eur J Haematol. 1998; 61(1): 59-64.
10.1111/j.1600-0609.1998.tb01062.x
CAS PubMed Web of Science® Google Scholar
7Ponce F, Magnol JP, Ledieu D, et al. Prognostic significance of morphological subtypes in canine malignant lymphomas during chemotherapy. Vet J. 2004; 167(2): 158-166.
10.1016/j.tvjl.2003.10.009
PubMed Web of Science® Google Scholar
8Miller AG, Morley PS, Rao S, Avery AC, Lana SE, Olver CS. Anemia is associated with decreased survival time in dogs with lymphoma. J Vet Intern Med. 2009; 23(1): 116-122.
10.1111/j.1939-1676.2008.0210.x
CAS PubMed Web of Science® Google Scholar
9Marconato L, Stefanello D, Valenti P, et al. Predictors of long-term survival in dogs with high-grade multicentric lymphoma. J Am Vet Med Assoc. 2011; 238(4): 480-485.
10.2460/javma.238.4.480
PubMed Web of Science® Google Scholar
10Kiupel M, Teske E, Bostock D. Prognostic factors for treated canine malignant lymphoma. Vet Pathol. 1999; 36(4): 292-300.
10.1354/vp.36-4-292
CAS PubMed Web of Science® Google Scholar
11Rebhun RB, Lana SE, Ehrhart EJ, Charles JB, Thamm DH. Comparative analysis of survivin expression in untreated and relapsed canine lymphoma. J Vet Intern Med. 2008; 22(4): 989-995.
10.1111/j.1939-1676.2008.0143.x
CAS PubMed Web of Science® Google Scholar
12Rao S, Lana S, Eickhoff J, et al. Class II major histocompatibility complex expression and cell Size independently predict survival in canine B-cell lymphoma. J Vet Intern Med. 2011; 25(5): 1097-1105.
10.1111/j.1939-1676.2011.0767.x
CAS PubMed Web of Science® Google Scholar
13Yamazaki J, Takahashi M, Setoguchi A, Fujino Y, Ohno K, Tsujimoto H. Monitoring of minimal residual disease (MRD) after multidrug chemotherapy and its correlation to outcome in dogs with lymphoma: A proof-of-concept pilot study. J Vet Intern Med. 2010; 24(4): 897-903.
10.1111/j.1939-1676.2010.0536.x
CAS PubMed Web of Science® Google Scholar
14Barber LG, Weishaar KM. Criteria for designation of clinical substage in canine lymphoma: a survey of veterinary oncologists. Vet Comp Oncol. 2016; 14: 32-39.
10.1111/vco.12086
PubMed Web of Science® Google Scholar
15Keller ET, MacEwen EG, Rosenthal RC, Helfand SC, Fox LE. Evaluation of prognostic factors and sequential combination chemotherapy with doxorubicin for canine lymphoma. J Vet Intern Med. 1993; 7(5): 289-295.
10.1111/j.1939-1676.1993.tb01021.x
CAS PubMed Web of Science® Google Scholar
16Jagielski D, Lechowski R, Hoffmann-Jagielska M, Winiarczyk S. A retrospective study of the incidence and prognostic factors of multicentric lymphoma in dogs (1998–2000). J Vet Med A Physiol Pathol Clin Med. 2002; 49(8): 419-424.
10.1046/j.1439-0442.2002.00458.x
CAS PubMed Web of Science® Google Scholar
17Sorenmo K, Overley B, Krick E, Ferrara T, LaBlanc A, Shofer F. Outcome and toxicity associated with a dose-intensified, maintenance-free CHOP-based chemotherapy protocol in canine lymphoma: 130 cases. Vet Comp Oncol. 2010; 8(3): 196-208.
10.1111/j.1476-5829.2010.00222.x
CAS PubMed Web of Science® Google Scholar
18Moore LD, Le T, Fan G. DNA methylation and its basic function. Neuropsychopharmacology. 2013; 38(1): 23-38.
10.1038/npp.2012.112
CAS PubMed Web of Science® Google Scholar
19Siegel RL, Miller KD, Jemal A. Cancer statistics. CA Cancer J Clin. 2019; 69(1): 7-34.
10.3322/caac.21551
PubMed Web of Science® Google Scholar
20Zainuddin N, Kanduri M, Berglund M, et al. Quantitative evaluation of p16(INK4a) promoter methylation using pyrosequencing in de novo diffuse large B-cell lymphoma. Leuk Res. 2011; 35(4): 438-443.
10.1016/j.leukres.2010.10.001
CAS PubMed Web of Science® Google Scholar
21Guan H, Xie L, Leithäuser F, et al. KLF4 is a tumor suppressor in B-cell non-Hodgkin lymphoma and in classic Hodgkin lymphoma. Blood. 2010; 116(9): 1469-1478.
10.1182/blood-2009-12-256446
CAS PubMed Web of Science® Google Scholar
22Chambwe N, Kormaksson M, Geng H, et al. Variability in DNA methylation defines novel epigenetic subgroups of DLBCL associated with different clinical outcomes. Blood. 2014; 123(11): 1699.
10.1182/blood-2013-07-509885
CAS PubMed Web of Science® Google Scholar
23Ferraresso S, Aricò A, Sanavia T, et al. DNA methylation profiling reveals common signatures of tumorigenesis and defines epigenetic prognostic subtypes of canine diffuse large B-cell lymphoma. Sci Rep. 2017; 7(1): 1-11.
10.1038/s41598-017-11724-w
CAS PubMed Google Scholar
24Aresu L, Ferraresso S, Marconato L, et al. New molecular and therapeutic insights into canine diffuse large B-cell lymphoma elucidates the role of the dog as a model for human disease. Haematologica. 2019; 104(6):e256.
10.3324/haematol.2018.207027
CAS PubMed Web of Science® Google Scholar
25Hsu CH, Tomiyasu H, Lee JJ, et al. Genome-wide DNA methylation analysis using MethylCap-seq in canine high-grade B-cell lymphoma. J Leukoc Biol. 2021; 109(6): 1089-1103.
10.1002/JLB.2A0820-673R
CAS PubMed Web of Science® Google Scholar
26Sato M, Mochizuki H, Goto-Koshino Y, et al. Prognostic significance of hypermethylation of death-associated protein kinase (DAPK) gene CpG island in dogs with high-grade B-cell lymphoma. Vet Comp Oncol. 2018; 16(3): 409-415.
10.1111/vco.12395
CAS PubMed Web of Science® Google Scholar
27Valli VE, Myint M, Barthel A, et al. Classification of canine malignant lymphomas according to the world health organization criteria. Vet Pathol. 2011; 48(1): 198-211.
10.1177/0300985810379428
CAS PubMed Web of Science® Google Scholar
28Yamazaki J, Jelinek J, Hisamoto S, Tsukamoto A, Inaba M. Dynamic changes in DNA methylation patterns in canine lymphoma cell lines demonstrated by genome-wide quantitative DNA methylation analysis. Vet J. 2018; 231: 48-54.
10.1016/j.tvjl.2017.11.007
CAS PubMed Web of Science® Google Scholar
29Tamura K, Yagihara H, Isotani M, Ono K, Washizu T, Bonkobara M. Development of the polymerase chain reaction assay based on the canine genome database for detection of monoclonality in B cell lymphoma. Vet Immunol Immunopathol. 2006; 110(1-2): 163-167.
10.1016/j.vetimm.2005.10.009
CAS PubMed Web of Science® Google Scholar
30Yagihara H, Tamura K, Isotani M, Ono K, Washizu T, Bonkobara M. Genomic organization of the T-cell receptor gamma gene and PCR detection of its clonal rearrangement in canine T-cell lymphoma/leukemia. Vet Immunol Immunopathol. 2007; 115(3–4): 375-382.
10.1016/j.vetimm.2006.11.005
CAS PubMed Web of Science® Google Scholar
31Ishizaki T, Yamazaki J, Jelinek J, Aoshima K, Kimura T. Genome-wide DNA methylation analysis identifies promoter hypermethylation in canine malignant melanoma. Res Vet Sci. 2020; 132: 521-526.
10.1016/j.rvsc.2020.08.006
CAS PubMed Web of Science® Google Scholar
32Yamazaki J, Meagawa S, Jelinek J, et al. Obese status is associated with accelerated DNA methylation change in peripheral blood of senior dogs. Res Vet Sci. 2021; 139: 193-199.
10.1016/j.rvsc.2021.07.024
CAS PubMed Web of Science® Google Scholar
33Ohta H, Yamazaki J, Jelinek J, et al. Genome-wide DNA methylation analysis in canine gastrointestinal lymphoma. J Vet Med Sci. 2020; 82(5): 632-638.
10.1292/jvms.19-0547
CAS PubMed Web of Science® Google Scholar
34Licenziato L, Minoli L, Ala U, et al. Predicting outcome in dogs with diffuse large B-cell lymphoma with a novel immune landscape signature. Vet Pathol. 2023; 60(3): 308-315.
10.1177/03009858231162209
CAS PubMed Web of Science® Google Scholar
35Marconato L, Aresu L, Stefanello D, et al. Opportunities and challenges of active immunotherapy in dogs with B-cell lymphoma: a 5-year experience in two veterinary oncology centers. J Immunother Cancer. 2019; 7(1): 146.
10.1186/s40425-019-0624-y
PubMed Web of Science® Google Scholar
36Dias JNR, André AS, Aguiar SI, Gil S, Tavares L, Aires-da-Silva F. Immunotherapeutic strategies for canine lymphoma: Changing the odds against non-Hodgkin lymphoma. Front Vet Sci. 2021; 8: 961.
10.3389/fvets.2021.621758
Web of Science® Google Scholar
37Saba CF, Clifford C, Burgess K, et al. Rabacfosadine for naïve canine intermediate to large cell lymphoma: Efficacy and adverse event profile across three prospective clinical trials. Vet Comp Oncol. 2020; 18(4): 763.
10.1111/vco.12605
CAS PubMed Web of Science® Google Scholar
38Marconato L, Frayssinet P, Rouquet N, et al. Randomized, placebo-controlled, double-blinded chemoimmunotherapy clinical trial in a pet dog model of diffuse large B-cell lymphoma. Clin Cancer Res. 2014; 20(3): 668-677.
10.1158/1078-0432.CCR-13-2283
CAS PubMed Web of Science® Google Scholar

Volume38, Issue1

January/February 2024

Pages 316-325

Use of genome-wide DNA methylation analysis to identify prognostic CpG site markers associated with longer survival time in dogs with multicentric high-grade B-cell lymphoma