Presence of herpesviruses, parvoviruses, and polyomaviruses in sinonasal lymphoma

The research ethics committee at the Helsinki University Hospital approved the study (§31/07.03.2019) and a research permission was granted (HUS/332/2019).

This was a retrospective cross-sectional study. The patients were treated at the Departments of Otorhinolaryngology—Head and Neck Surgery and Oncology, Helsinki University Hospital, Helsinki, Finland. We recorded the clinical information from hospital charts (Table 1).

We collected FFPE tumours representing 24 SL patients diagnosed during 1987–2018 from the Helsinki Biobank (permission no §73/15.05.2019, HUS/118/2019). The Helsinki Biobank provided plasma samples from seven SL patients with available FFPE tumour samples and additionally from three SL patients with no tumour samples available.

We analysed all tumours by qPCR for viral DNA, and the high-load ones by RNAscope in situ hybridisation (RISH) for cell tropism. We collected plasma samples to investigate the serological status of parvoviruses B19V, BuV, CuV and TuV, and one underwent targeted NGS.

All tumour biopsies were collected in a PCR-sterile manner from FFPE tissue blocks as 2-mm punch biopsies in sterile 1.5 ml microcentrifuge tubes. The correct localization was evaluated microscopically from the representative tissue slide by the first author, and whenever in doubt, confirmed by the senior pathologist. If enough material were available, two distinct biopsies from the tumour were taken. Prior to puncturing, new HE slices were cut from paraffin blocks to ensure the punching to hit tumour tissue. DNA was extracted with QIAamp DNA FFPE Tissue Kit (Qiagen, Heiden, Germany), according to the manufacturer’s protocol, with slight modifications: the paraffin treatment with xylene was done twice and 40 µl of proteinase K was used. We eluted the DNA preps in 60 µl AE buffer and stored them at -20°C. From one plasma sample (patient L31), we extracted DNA with QIAamp DNA blood Mini Kit (Qiagen), according to the manufacturer’s protocol. The DNA yields and human cell quantity were evaluated by comparing viral loads with that of the human reference gene RNase P by qPCR [25].

The viruses analysed with qPCR included the herpesviruses (herpes simplex virus-1 and -2, varicella zoster virus, EBV, cytomegalovirus (CMV), HHV-6A, HHV-6B, HHV-7, and HHV-8), the polyomaviruses (BK polyomavirus (BKPyV), JC polyomavirus (JCPyV), and MCPyV), and the parvoviruses (B19V, BuV, CuV, TuV, and HBoV1–4). Details are listed in Table 2 [14, 25‐28]. All qPCR reactions contained either Maxima probe qPCR Master Mix (Thermo Fischer Scientific, Pittsburgh, PA, USA) or TaqPath ProAmp master mix Thermo Fisher), as described [29]. We performed all real-time qPCR assays with AriaMx Realtime PCR System (Agilent Technologies, Santa Clara, CA, USA).

We included molecular biology grade water in all qPCR reactions as a non-template control and used ten-fold diluted plasmids (10¹–10⁶), containing each viral target regions as qPCR standards and as positive controls.

We processed one plasma sample of patient L31, in whose tissue sample we had detected high copy numbers of CuV DNA, by targeted NGS, given the unknown nature of this rather newly discovered virus. The viruses targeted with the NGS are listed in Table 2. We fragmented the DNA mechanically and prepared the libraries with unique double indexes and performed in solution capture-targeted enrichment of the viral DNAs [29, 30]. The enriched library was sequenced, at the FIMM genomics unit of the University of Helsinki, with Novaseq 6000 (S1, PE151 kit; Illumina). The pipeline was validated as described in Pratas et al. [31] and has been tested on a wide range of materials, including human tissues, bone, serum/plasma, and formalin-fixed samples [24]. A blank extraction control was carried downstream alongside the sample throughout the entire process, spanning library preparation, enrichment, and sequencing.

The data were analysed with TRACESPipeLite, a fully automatic program [30‐32]. When in low breadth coverage (< 15%), we manually inspected and confirmed the individual reads by BLAST. The viral reads of cutavirus were manually verified using BLAST and confirmed to match exclusively cutavirus.

We chose five tumours for RISH; (Advanced Cell Diagnostics (ACD), Newark, CA), each representing the highest virus DNA load (3.45 × 10²–2.08 × 10⁶ cpm) of either CuV, EBV, HHV-6B, HHV-7 or B19V, and five additional virus DNA-negative SL tumours. We performed RISH for the CuV PCR-positive tumour of patient L31, the EBV PCR-positive tumour of patient L28, the HHV-6B and -7 PCR-positive tumour of patient L35, and the B19V PCR-positive tumour of patient L36. We applied RISH on 5-µm thick FFPE sections on SuperFrost glass slides with RNAscope 2.0 HD Red Chromogenic Reagent Kit (ACD) with probes targeting the regions BILF-1 (catalogue Number 44781) of EBV, U4 (870,801) of HHV-7; U36 (521,401) of HHV-6B, NS (869,391) and VP1 (872,961) of CuV and NS1 (496,871) of B19V, according to the manufacturer’s protocol. The human housekeeping gene PPIB and the bacterial gene DapB (ACD) served as positive and negative technical controls, respectively.

To identify the host cells of CuV in the tumour of patient L31, we combined RNAscope in-situ hybridization (RISH) with fluorescent multiplex immunohistochemistry (mIHC) on 5-mm FFPE tissue sections, with multiple cellular markers on the same tissue section. The CuV probe (Probe-V-CuV-gp3-gp4-gp5 [NC_039050.1 (nt 2023-3562)], targets the sense DNA strand and mRNA of the VP gene region [15]. Probes targeting the human housekeeping gene PPIB and the bacterial gene DapB (ACD) served as positive and negative technical controls, respectively.

RISH (RNAscope 2.5 HD reagent kit-BROWN; [PN 322310; ACD, Newark, USA]) was employed according to the manufacturer’s protocol with slight modification. After addition of the AMP6 reagent, Alexa Fluor™ 488 Tyramide Reagent (Thermo Fisher [cat: B40953]) was applied to the tissue sections in a 1:100 dilution for 5 min, to make the RISH signal fluorescent. This was then followed by mIHC, as previously described (Blom et al., 2017), with some modifications. Briefly, primary antibodies raised in different species were applied pairwise in sequential staining rounds and detected by AlexaFluor647 and AlexaFluor750 fluorochrome-conjugated secondary antibodies. The following primary biomarker antibodies (1:200): helper T cells, (Rabbit-anti-CD4; Abcam; ab133616), cytotoxic T cells (Mouse-anti-CD8; Dako; M7103), macrophages (Mouse-anti-CD163; Thermo; MS-340) and B cells (Rabbit-anti-CD163; Abcam ab188571). DAPI was applied as a nuclear stain. After each round of staining the tissues were scanned on a Zeiss Axioscan.Z1 slide scanner. Before the next round of staining, coverslips were soaked off in wash buffer and the tissues were incubated in a bleaching solution (TBS/24 mM NaOH/4.5% H₂O₂) for 1 h at room temperature. Then the slides were heated in 10 mM Tris–HCl/1 mM EDTA, pH 9, for 20 min at 99°C to inactivate the primary antibodies of the previous staining round. Representative CuV RISH-mIHC- and RISH only-treated SL biopsy tissues are shown in Fig. 1.

Seven of the 24 patients included in this study, had plasma samples stored at the Helsinki Biobank. Plasma samples but no FFPE tumour blocks were available from three additional patients with SL. We tested all the 10 plasma samples by in-house IgG EIAs, using biotinylated VP2 virus-like particles (VLP) as antigen for BuV, CuV and TuV, as well as for B19V, as previously described [33, 34]. The initial results for protoparvovirus IgGs were confirmed with a competition assay [33]. Optical densities (ODs) were measured at 450 nm (Multiskan EX; Thermo Fischer Scientific).

Clinical information was available for 17 patients and unavailable for seven patients (Table 1). Overall, 50% of the patients were female. All the tumours were known to be situated in the sinonasal area, but the exact location of the tumour was not available for eight patients. In addition to SL DLBCL-NOS (NOS—not otherwise specified), patient L18 had a systemic T-cell lymphoma. Pathological reports were available for all patients, but in some cases the diagnosis did not follow the latest lymphoma classification due to old age of samples. The histological SL subtypes represented DLBCL-NOS (n = 16, 67%), mantle-cell lymphoma (MCL, n = 2, 8%), and smaller entities (Table 1). Immunostainings for EBV and ALK (anaplastic lymphoma kinase positive) were reported for some samples: samples L19 and L21 were EBNA2 and LMP1 negative, L25 was ALK unknown, L28 was positive and L34 was negative for EBER by chromogenic in situ hybridisation. It is noteworthy that there were no lymphomas of NK-cell subtype within this cohort.

The cohort included many patients with compromised immunity before the manifestation of SL (Table 1). Reasons such as malnutrition or tobacco smoking, were not reported in the patient charts.

We detected virus DNA in 15/24 (63%) tumours with viral loads ranging between 4.4 and 2.1 × 10⁶ copies per one million cells (cpm) (Table 3). Nine tumours were positive for EBV DNA (38%; 3.6 × 10¹–2.1 × 10⁶ cpm). The DLBCL-NOS subtype harboured EBV DNA in 6/16 (38%) of the tumours. Both MCLs were EBV-DNA positive as well as the single anaplastic tumour.

Six tumours were positive for HHV-7 DNA (25%; 2.9 × 10¹–4.8 × 10² cpm), detected in almost all subtypes present in our study: DLBCL-NOS, MCL, and anaplastic lymphoma. We found B19V DNA and HHV-6B DNA in four tumours each (17%; 2.9 × 10¹–2.1 × 10³ cpm; 4.4 × 10¹–3.5 × 10² cpm, respectively): B19V DNA in DLBCL-NOS, and HHV-6B in MCL, and DLBCL-NOS. We found CMV DNA in two tumours (8%; 1.6 × 10²–8.7 × 10² cpm) representing MCL, and DLBCL-NOS, respectively; and detected CuV DNA (3.1 × 10⁵ cpm) and MCPyV DNA (2.1 × 10² cpm) in one tumour each (4%): both DLBCL-NOS. The human single-copy RNase-P qPCR served for quantification of human cells with a variation of 2.3 × 10³–1 × 10⁶ (mean: 2.7 × 10⁴) cells/ul DNA extract. All positive findings were reproducible.

Two or more different viral DNAs were co-detected in 8/24 (33%) tumours with up to four different viruses per tumour (Table 3). EBV was co-detected with other viruses in 78% of EBV-positive tumour samples, most often with HHV-7. HHV-6B, -7, and CMV were never the only viral findings in tumours but were present in combinations with other viruses; with EBV in 22%, 44%, and 100%; with B19V in 17%, 25%, and 0%; and with MCPyV in 0%, 25%, and 0% of HHV-6B, HHV-7 or CMV-positive  SL samples, respectively.

Among the five patients that were immunocompromised before the onset of the lymphoma, we detected virus DNA in four. Two out of the three patients with HIV were EBV-DNA positive. A slight emphasis was noted in the overall virus DNA prevalence (63% vs 80%) and the co-detections (1.6 virus co-detections/tumour vs 1.8 co-detections/tumour), between our groups of healthy, and the previously immunocompromised individuals, respectively.

While the mean age of the  patients at the time of SL diagnosis was 66 years (range 34–86), the mean age at diagnosis for the virus DNA-positive and -negative groups was 62 years (SD ± 14, range 34–84) and 73 years (SD ± 12, range 46–86), respectively. However, the age difference was not statistically significant (p = 0.06).

As the relevance of CuV persistence is unknown and interesting, we analysed the plasma sample from the patient with CuV PCR-positive SL (SL31) with targeted NGS to investigate a possible viremia (Table 3). The plasma sample was NGS positive for CuV (seven unique reads). We could, however, not repeat the result of plasma CuV positivity with qPCR.

To investigate EBV, we chose a high copy-number DLBCL-NOS tumour (patient L28, AIDS positive, Tables 1, 3). The EBV probe BILF-1 demonstrated scattered signals on a broad area of the tumour (Fig. 1d). In the CuV-positive tumour (patient L31), we detected repeatedly intense positive signals by both CuV probes VP1 and NS1, matching with each other on the layout, with signals in single spots or small clusters spread out unevenly throughout the tumour (Fig. 1 a-c).

When combining RISH with IHC on the same slide, we could localize CuV in CD4+ T cells, however, with fewer recognizable positive CuV spots when compared to RISH alone (Fig. 1a–c and i–l). We did not detect CD8+ , B cells or macrophages overlapping or in direct contact with CuV signals.

To detect HHV-6B and -7, and B19V, we chose tumour samples with the highest, however still modest, virus loads (2.07 × 10² cpm; patient L35 for both HHVs, 2.07 × 10³ cpm for B19V), without solid findings. PCR-negative tumour controls remained all negative.

To search for signs of immune system activation, and to survey the seroprevalence of the ‘new’ protoparvoviruses, we performed protoparvovirus and B19V EIAs for all the 10 plasma samples available. Two out of the 10 plasma samples were positive for CuV-specific IgG (absorbance values 1.5 and 2.1). One of the CuV IgG-positive plasma samples belonged to patient L31, who was positive for CuV DNA both in FFPE tumour (qPCR) and in plasma (NGS). The other CuV IgG-positive plasma belonged to a SL patient with no FFPE-tumour sample available. One plasma sample was BuV2-IgG positive. The IgG prevalence of B19V was 60% (6/10).