L929 (mouse fibroblast; Oncolytics Biotech) were cultured in DMEM. Media was supplemented with 5% (v/v) Fetal Calf Serum (FCS) (or 2% (v/v) FCS for plating media), 1% (v/v) glutamine, and 0.5% (v/v) penicillin/streptomycin.

Mammalian orthoreovirus type 3 Dearing (Pelareorep) in PBS was obtained from Oncolytics Biotech (Calgary, AB, Canada) and stored at −80°C. Viral titer was regularly confirmed by TCID₅₀ assay.

Samples were collected from patients at various time points within their treatment course according to the protocol-defined schedules. Clotted blood samples were taken and within 4 hours, centrifuged at 3000 rpm for 10 min at room temperature and stored at −80°C. Patient sera were heat inactivated at 56°C for 30 min prior to antibody analysis. To calculate patients’ antibody titers, a modified neutralizing antibody assay was employed as described previously.4 Briefly, a known titer of reovirus was incubated with serial dilutions of patients’ serum and the neutralization of virus particles was assessed by loss of ability to kill a monolayer of L929 cells (analyzed by the MTT assay). A goat polyclonal antibody was used as a positive control.

For accuracy, two methods of calculating antibody titers were employed. The highest dilution that displayed neutralization (>80% cell killing) in any of the replicate wells was classed as the endpoint titer. The second method of calculating antibody titer used the highest dilution of patient serum that gave >50% neutralization in the replicate wells (eg, <80% cell killing in over half of replicate wells); this is classed as the halfpoint titer.

To quantify changes in patient serum antibody isotype titers in response to reoviral administration, high-throughput 7-plex antibody isotyping magnetic bead-based Luminex assays (Thermo Fisher) were performed according to manufacturer’s instructions. Serum concentrations of IgG1, IgG2, IgG3, IgG4, IgM, IgA and IgE were quantified at defined time-points: pretreatment (0), 5, 8 and 15 days following the initiation of each dosage cycle. Serum was stored at −80°C and thawed on ice prior to analysis.

Antibody data were collected from various phase I/II clinical trials administering RT3D intravenously as a single agent,8 in combination with gemcitabine,16 docetaxel,17 in combination with the two chemotherapeutic agents paclitaxel and carboplatin,15 and in combination with cyclophosphamide,21 in patients with advanced malignancies.

Low sample number for some cohorts at individual time points reduced the power of statistical comparisons at those time points. To overcome this, where noted, the area under the curve from baseline to C2D1 was calculated for each cohort to create comparisons of NARA response over the first 2 weeks of treatment that can be subjected to statistical analysis. Kruskal-Wallis non-parametric tests, and two-sided paired t-tests were used to compare between and within cohorts, respectively. Data are presented as mean±SEM, and p values were reported as: ns=not significant, p<0.05 (*) and p<0.01 (**). All statistical analysis was performed using GraphPad Prism V.7.0.

Details of patients who were enrolled in a series of clinical trials of reovirus, with or without different chemotherapies, and the relevant treatment and sampling schedules are summarized in figure 1. The different combinations allowed us to address the effect of multiple chemotherapy drugs on the NARA response in patients.

Antibody levels against reovirus were assessed by a modified neutralizing antibody assay as described previously.4 Individual patients’ NARA responses to intravenous reovirus monotherapy (REO005 trial) are shown in figure 2A(i). Despite heat inactivation of complement, all patients’ serum displayed a limited pre-existing ability to neutralize reovirus prior to reovirus treatment, although this baseline was variable, consistent with the known exposure to reovirus of the general population.22 23 NARA titers generally began to increase 5–8 days after the first reovirus infusion. Patients most often reached peak titer around 15 days postinfusion.

For patients who completed at least two cycles of treatment and where an end-of-study sample was available, NARA endpoint titers pretreatment, at peak and at the end-of-study are summarized in figure 2A(i) and show that peak titers are maintained at or near the peak until end of study for most patients, up to 54 days following the last infusion. Complete endpoint antibody titers at all measured time points in REO005 are summarized in online supplemental figure 1A–H.

Individual patients’ NARA responses, when reovirus was given in combination with gemcitabine (REO009), docetaxel (REO010), carboplatin plus paclitaxel (REO011) and cyclophosphamide (REO012) are shown in figure 2B. Patients’ NARA responses appear similar to those seen with reovirus monotherapy, peak titers being reached around day 15; however, the peak titers themselves were suppressed with gemcitabine treatment, being 1–2 logs lower than in the other trials. This apparent difference may be due to lower total doses of reovirus (one dose vs five consecutive doses), rather than due to concomitant gemcitabine treatment.

For patients treated with reovirus alone (REO005), average NARA titers are shown per cohort in figure 3A. Most patients receiving reovirus alone (REO005) appeared to develop a marked antibody response by C1D8 and peak levels of antireovirus antibody levels within their first cycle of treatment by days 8–15. There was no significant increase in peak NARA titers associated with increasing reovirus dose.

Figure 3B shows the same information for patients receiving reovirus plus chemotherapy. Again, no significant increases in peak NARA titers were observed in response to increasing reovirus doses for any drug combination. NARA responses appear to be delayed in combinations with docetaxel or gemcitabine (but less so with carboplatin/paclitaxel), as illustrated by the longer time taken for titers to approach peak levels (compare with data in figure 3A). This suggests that there may be a longer window at the start of certain combination studies in which to ‘load’ virus before antiviral antibody levels rise too high.

It is important to note that a reduced reovirus dose frequency was used for the patients receiving gemcitabine because of observed hepatic and cardiac toxicities within the first two patients (excluded from all analyses here because of incomplete acquisition of data). It is also important to note that viral replication was not implicated directly in these toxicities and the virus remained non-pathogenic. However, reovirus dosing was reduced from the standard five doses (days 1–5) per cycle used in the other trials to one dose (on day 1), and treatment-related toxicities were not seen with this regimen. It is possible that this dose reduction was responsible for the lower NARA titers, and to address this further, the ‘area under the curve’ (AUC) of antibody levels, from baseline to C2D1, was calculated for each patient in cohorts 1, 2 and 3 of the single agent REO005 trial who received a dose of 1×10⁸ TCID₅₀ once, thrice or on five consecutive days, respectively. This escalating frequency of reovirus dosing had no significant effect on the AUC NARA response in the early cohorts of REO005, although only one data point was available for the single infusion cohort, as dictated by the trial design (figure 3C). Additionally, the higher 1×10⁹ TCID₅₀ dose used with gemcitabine in cohort 2 of REO009 shows extensive suppression of the NARA response, with an area under the curve several logs lower than all the reovirus alone cohorts, despite the higher total virus dose. In addition, the titer remains close to the baseline level until day 15 (online supplemental figure 3A). This analysis shows that the suppression of the NARA response observed with gemcitabine was likely a genuine drug effect, rather than a dose-frequency effect. Importantly, after the number of doses of reovirus in this regimen was reduced from 5 to 1 (and all data presented here reflect this dose reduction), the combination of reovirus and gemcitabine was well tolerated.

Because REO005, REO010 and REO011 used similar doses and treatment regimens, it was possible directly to compare NARA responses to equivalent viral doses in the presence or absence of chemotherapy. Cohort 1 in REO010 and in REO011 and cohort 6 in REO005 all received reovirus at the same dose of 3×10⁹ TCID₅₀, on five consecutive days; cohort 3 of REO009 received a single dose of 3×10⁹ TCID₅₀. This allowed a direct comparison of the effects on NARA titers of docetaxel, carboplatin/paclitaxel doublet and a close comparison for gemcitabine. Likewise, cohort 5 from REO009 and cohort 2 from REO010 and from REO011 can be compared with cohort 7 in REO005 to make the same comparison for patients receiving 1×10¹⁰ TCID₅₀ doses (refer to figure 1A).

Again, area under the curve to C2D1 was calculated for these comparisons, showing that at the 3×10⁹ dose (figure 3D), the NARA titer was significantly suppressed by gemcitabine and docetaxel but not by carboplatin/paclitaxel. Although docetaxel was suppressive, this effect was short lived, and the peak titer was still reached by C1D15, at a comparable level to titers for patients receiving reovirus alone. In contrast, gemcitabine suppression lasts longer and reduced the final peak titer approximately 10-fold (online supplemental figure 3B). At the 1×10¹⁰ dose, significant suppressive effects were seen again (figure 3E) and for longer duration, with suppression in the titer at C1D15 still apparent for both gemcitabine and docetaxel (online supplemental figure 3C). Again, no suppressive effect of carboplatin/paclitaxel was seen.

The immunosuppressive drug cyclophosphamide was given in increasing doses in combination with a fixed dose of reovirus (3×10¹⁰ TCID₅₀) in the trial REO012, specifically to find a dose that would delay the NARA response, theoretically allowing enhanced tumor colonization and replication of the virus. The NARA response to 3×10¹⁰ TCID₅₀ concomitantly with increasing doses of cyclophosphamide is shown in figure 4A. No clear cyclophosphamide dose effect was apparent in these NARA titers. As previously, we compared these data with cohorts receiving similar reovirus doses in the other trials, that is, cohort 8 from REO005, cohort 5 from REO009 and cohort 3 from both REO010 and REO011, compared with the cohort with the highest dose of cyclophosphamide (1000 mg), cohort 9 from REO012. Compared with the other chemotherapy agents at this dose of reovirus, cyclophosphamide appears to slow the initial NARA response, with titers lower at C1D8 than for all comparators (figure 4B). Analysis of area under the curve to C2D1 shows that half of the patients appear to have had suppressed NARA responses, but there was no overall statistical significance (figure 4C). Notably, the suppression seen previously for gemcitabine and docetaxel remains clear at this higher dose of virus.

Given that the effect of escalating virus dose on NARA responses in each trial was minimal, it was possible to pool all the patients in each trial together to get a comprehensive view of how NARA titers were affected by the drug combinations. We assessed whether all patients on each trial, regardless of virus dose, reached peak titer during or after the first cycle of treatment (figure 4D). This showed that the most effective drug for delaying the peak NARA titer is gemcitabine, where 80% of patients reached peak NARA response in the later cycle. Cyclophosphamide was the second most effective but resulted in a later peak titer for only half of patients. To normalize for the potential effect of varying baseline titers, NARA titer fold change over time was analyzed for all patients in each trial (figure 4E). This shows that docetaxel and cyclophosphamide have a similar suppressive profile at C1D8 but that this was lost by C1D15, from which point the levels were similar to reovirus alone. Carboplatin/paclitaxel was the only combination that shows a dip in the antibody titer before the second cycle, and gemcitabine displays the slowest and overall lowest increase in titers.

To explore the NARA response in more detail, the total titers of antibody subtypes in blood sera were quantified to determine their individual kinetics in response to reovirus (figure 5). This quantification included all antibodies in circulation, not just reovirus-specific antibodies, but it was assumed that responses seen in the window of treatment were primarily antireovirus. Patients receiving 1×10¹⁰ TCID₅₀ were assessed for IgG1, IgG2, IgG3, IgG4, IgM, IgA and IgE in blood serum. The kinetics of each isotype were broadly similar, all initially peaking at day 8 postinfusion before reverting to pretreatment levels before the second infusion (figure 5A). A response was seen in all isotypes, but the highest mean titer (across n=5 patients for each timepoint) was for IgG1 (10.3 mg/mL at C1D8), followed by IgM (3.2 mg/mL at C1D8), and the remaining isotypes approximately 1 log lower. Despite all isotype titers increasing in response to virus treatment, the IgG1 response was so overwhelmingly large that the percentage proportions of other isotypes were depressed in comparison (figure 5B). As a representative example, the kinetics of individual isotypes from a single patient were compared (figure 5C), clearly showing the overwhelming IgG1 response. Despite the total antibody titers returning to pretreatment levels between cycles of treatment, the NARA levels remained at peak titer from day 8 onwards (figure 5D). This indicates that the vast majority of antibodies produced, even if reovirus-specific, were not able to neutralize the virus, as the NARA titers do not decrease at all even when the total antibody titers return to pretreatment levels.

The NARA data showed different levels and duration of NARA suppression by gemcitabine and docetaxel. To determine if these differences were due to isotype switching from IgG1 to a less effective isotype such as IgG4, sera from patients receiving the same 1×10¹⁰ TCID₅₀ dose in combination with these drugs were also analyzed (figure 6).

The significant concentration increases seen in every isotype (except IgG4) between baseline and C1D8 were completely suppressed by both gemcitabine and docetaxel, with levels unchanged compared with baseline in each case. Total immunoglobulin concentration was significantly reduced by gemcitabine, but no significant change was seen with docetaxel. This observation is consistent with gemcitabine having a larger effect on the NARA responses observed.

The 1×10¹⁰ TCID₅₀ dose was able to stimulate a broad range of isotypes, with all showing a significant increase in concentration except IgG4. Lower doses were also compared as the gemcitabine was given with just a single infusion of virus. These ranged from 1×10⁸ TCID₅₀ given once, 1×10⁸ TCID₅₀ given five times and 1×10⁹ TCID₅₀ also given five times. These doses were grouped as R^low in figure 6 and appear to show a reduced IgG1 response compared with the higher dose (R^high) as well as no significant IgG3 or IgM response, altogether resulting in no significant change in the total immunoglobulin concentration.