Early salivary changes in multiple myeloma patients undergoing autologous HSCT

Objective : One explorative observational study in two parts was performed to examine early salivary changes in relation to oral mucositis (OM) in multiple myeloma patients treated with high- dose melphalan and autologous haematopoietic stem cell transplantation (HSCT). As cryotherapy was introduced after part A as regular care, its effect on OM could be evaluated. Methods : Unstimulated whole- mouth saliva (UWS) and stimulated whole- mouth saliva (SWS) were collected, and OM was scored with the Oral Mucositis Nursing Instrument (OMNI) at days −3, 0, 4, 7, 11 and 14 after HSCT. Salivary flow rate, total protein (BCA), mucin 5B, albumin (western blot), total IgA, lactoferrin and myeloperoxidase levels (ELISA) were determined. Results : Trends of decreasing UWS and SWS flow rates and total IgA levels were observed. At days 7 and 11, increases in lactoferrin and albumin levels were found in UWS and SWS. A positive correlation was found between OMNI scores and albumin and lactoferrin levels in SWS ( R 2 = .56, p = .029 and R 2 = .49, p = .043, respectively). In part B, cryotherapy significantly lowered peak OMNI scores. Conclusion : Compositional changes in saliva reflecting inflammation were found in the first days after HSCT, and the use of cryotherapy in the second part was associated with decreased OM severity.


| INTRODUC TI ON
Haematopoietic stem cell transplantation (HSCT) is a potentially curative treatment for malignant and non-malignant diseases. An HSCT can be performed either with autologous or allogeneic stem cells. In an autologous HSCT, patient's own stem cells are collected, stored and reinfused after suppressing the immune system with high-dose chemotherapy. For example, high-dose melphalan (HDM) is the standard care treatment regimen for multiple myeloma (MM) (Voorhees & Usmani, 2016).
A common adverse effect of HDM is oral mucositis (OM), characterized by pain, erythema, swollen mucosa, xerostomia and ulcerations (Peterson & Cariello, 2004). Severe OM requires narcotics and parenteral feeding and increases hospital stay.
Furthermore, the ulcerative lesions may allow the penetration of bacteria into the bloodstream resulting in a systemic infection (Peterson & Cariello, 2004). Oral mucositis to some degree affects almost all patients treated with high-dose chemotherapy. A multicentre study to assess the incidence, severity and duration of OM found that 90% of the MM patients treated with | 973 van LEEUWEn Et aL. HDM and autologous HSCT was affected. Severe OM (WHO score 3 and 4) occurred in 46% of the MM patients and had a duration of 5 days (Blijlevens et al., 2008). Peak OM scores were found approximately 9 days after HSCT (Blijlevens et al., 2008). To prevent OM, cryotherapy is used for patients undergoing HDM and HSCT (Marchesi et al., 2017). Cryotherapy, the use of ice chips during the infusion of chemotherapy, results in local vasoconstriction, which reduces the amount of oncolytic drugs that reach the oral mucosa. Although it significantly reduces the incidence, some patients still develop severe OM (Marchesi et al., 2017). Moreover, there are no good treatment options for OM.
One factor that may influence the severity of OM is a decrease in the normal defence mechanisms in saliva (Epstein, Tsang, Warkentin, & Ship, 2002). Several salivary constituents are thought to have a protective function such as immunoglobulins, histatins, cystatins, defensins and lactoferrin (Amerongen & Veerman, 2002). A study evaluating changes in the salivary proteome after allogeneic HSCT found that levels of lactoferrin and secretory leucocyte protease inhibitor were increased at 6 months compared to pre-HSCT levels (Imanguli et al., 2007). The levels of secretory IgA (sIgA) were decreased 1 month after HSCT, but returned to pre-HSCT levels 6 months after HSCT (Imanguli et al., 2007). Decreased sIgA levels after HSCT were also found in another study examining salivary changes in MM patients undergoing autologous HSCT (Avivi et al., 2009). In the latter study, sIgA levels were significantly reduced 7 days after HSCT and OM was associated with a decrease in sIgA, uric acid levels and salivary antioxidant activity.
As OM is most severe around 9 days after transplantation, the aim of this study was to examine early salivary changes in relation to OM in MM patients treated with HDM and autologous HSCT. To determine salivary changes caused by leakage from the oral mucosae, unstimulated whole-mouth saliva (UWS) and stimulated (SWS) whole-mouth saliva were used in this explorative observational study performed in two parts. Due to protocol changes in regular care at the hospital based on the MASCC/ ISOO clinical practice guideline in 2014 for the prevention of OM, patients in the last part received cryotherapy during infusion of HDM (Lalla et al., 2014).

| MATERIAL S AND ME THODS
An explorative observational study in 2 parts was performed in 2013 (part A) and 2015 (part B). Approval was given by the Local Ethical Committee (registration number 2012/528), which stated that this research was not subjected to the law governing research involving human subjects and no approval of the Local Ethical Committee was necessary. The study was conducted in full accordance of the World Medical Association (2001). All patients were informed about the study and signed informed consent before participating in the study.

| Oral examination and saliva collection
During the hospitalization period, saliva was collected and the severity of OM was scored using the OMNI score (Oral Mucositis Nursing Instrument) on day −3 (admission day), day 0 (day of autologous HSCT), day 4, day 7, day 11 and day 14 (Potting, Blijlevens, Donnelly, Feuth, & van Achterberg, 2006). The total OMNI score was calculated by the sum of the individual scores (ranging between 0 and 3) for erythema, edema, ulcerations, pain, mouth dryness and saliva viscosity.
The protocols for the collection of UWS and SWS were based on the guidelines for saliva collection of the University of Southern California School of Dentistry (Navazesh & Kumar, 2008). For the collection of UWS and SWS, patients were asked to refrain from eating or drinking 1 hr before collection and to swallow immediately before collection. For UWS, patients were asked to spit the saliva in a previously weighed plastic cup for 5 min without making any effort to increase the saliva flow. For SWS, patients were asked to chew 1 min on a piece of paraffin film to stimulate saliva flow. After swallowing, SWS collection was started by spitting saliva, while continuing chewing, in a plastic cup for 2-5 min. Directly after collecting, samples were weighed and stored on ice. UWS and SWS flow rates were estimated by assuming 1 ml of saliva is 1 gram. Directly after collection, pH was determined using pH strips (pH-indicator strips 4.0 -7.0, Merck ® , Germany), and saliva samples were centrifuged for 5 min at 9,800 g (mini spin plus, Eppendorf, Germany). The supernatant was stored separately from pellets at −80°C. For analysis, samples were defrosted and divided into aliquots and stored at −80°C.

| Saliva analysis
Total protein was determined using the BCA protein assay kit according to the manufacturer's protocol (Pierce ™ BCA protein assay kit, Thermo Scientific, USA). Salivary proteins and glycoproteins were analysed by SDS-PAGE gels stained with Coomassie Brilliant Blue R250 (CBB) and Periodic Acid-Schiff (PAS) to determine mucin 5b (MUC5B). Western blotting was used to determine MUC5B and albumin levels, and enzyme-linked immunosorbent assay (ELISA) was used to determine total IgA, myeloperoxidase (MPO) and lactoferrin levels.

| SDS-PAGE and CBB/PAS staining
For SDS-PAGE gels, equal amounts of total protein were loaded onto gels, UWS and SWS separately. Samples were prepared in LDS containing sample buffer (1:4) and DTT (1:10), incubated at 95°C and centrifuged. Protein standards with predefined molecular weights (5 μl, SeeBlue ® Plus2 prestained protein, Thermo Scientific, USA) and 10 μl of sample were loaded on 4%-12% Bis-Tris gels (NuPage ™ , Thermo Scientific, USA) and run in MES SDS running buffer as per manufacturer's recommendations. Gels were stained in CBB staining solution (Brilliant Blue R Staining Solution in 10% acetic acid, Sigma-Aldrich, UK) and destained in 10% acetic acid. The CBB-stained gels (after photographing) were used for PAS staining after 1 hr incubation in fixing solution (25% methanol, 10% glacial acetic acid in H 2 O).
After incubation, gels were washed with H 2 O for 20 min, oxidized in 2% periodic acid for 15 min, washed with H 2 O twice and incubated in Schiff reagent (VWR ® Prolabo ® , UK). Images were taken and band intensities were determined using a Bio-Rad ® Chemidoc ™ Imaging system and Image Lab ™ software.

| Western blots
Using a sandwich technique, proteins were transferred from gel to a nitrocellulose membrane (Invitrogen, XCell II ™ Blot Module, USA) in transfer buffer (Novex ® NuPage Transfer Buffer 20x, USA, in methanol and H 2 O) for 1 hr at constant voltage. After transfer, the membrane was blocked in 5% skimmed milk in TBS-T. Blots were incubated with primary mouse monoclonal anti-human MUC5B (1:500 in TBS-T, Abcam, Ab105460, UK) or mouse anti-albumin (1:1,000 in TBS-T Sigma, A6684, UK). Blots were washed 3x 5 min with TBS-T and incubated with second antibody, polyclonal goat anti-mouse HRP (1:2,000 in TBS-T, Dako P0447, UK). Results were visualized using a chemiluminescence Western blotting reagent (Bio-Rad ® ClarityTM Western C Kit Western ECL Subrate, USA). Images were taken, and intensities were determined using Chemidoc and Image Lab ™ software.
After washing, 100 μl samples/standard was added in duplo to the plates and incubated at 37°C for 2 hr. IgA from colostrum I2636 (1 -0.0156 μg/ml in PBS-T, Sigma-Aldrich, UK), lactoferrin from human milk L0520 (

| Data analysis
Statistical analysis was performed in SPSS (version 22) and R (version 3.4.0), and p < .05 was considered as statistical significant.
Descriptive analysis was used for patient characteristics. To test the difference between the OMNI scores in part A and part B, a multilevel regression model was used with the OMNI score as dependent variable and "study part" as independent variable. To model the changes over time, both T and T 2 were included as independent variables as well, while a random intercept was added to the model to allow for the clustering of measurements within patients.
All quantitative results of MUC5B, albumin, total IgA, lactoferrin and MPO are based on part B only, as part A was used to optimize the protocols for quantitative analysis of saliva samples. Results from the salivary compositional analysis of the first part (A) were only used to compare trends between both studies. To correct for interindividual variation between patients, scaled scores were calculated per variable per patient by subtraction of day −3 score and dividing by the SD. These scaled scores were used for paired t tests to compare days 0, 4, 7, 11 and 14 with day −3. Differences are graphically labelled at α-levels of 5%, 1% and 0.1% separately to allow for evaluation of significance in view of multiple testing. For easier interpretation, most figures contain the absolute values. Correlations between peak OMNI scores and salivary parameters were calculated by applying meta-analytical techniques (inverse variance weighting) on the individual correlations within patients over time.

| Patient characteristics
In total, 20 patients were included (part A: n = 12, part B: n = 8). No important differences in study population were observed between both groups (Table 1). Due to hyposalivation or nausea, it was not possible to collect saliva from all patients on all study days. This resulted in missing samples with a peak at day 7 in both parts (Figure 1).
An important difference between both groups is the use of cryotherapy during the infusion of HDM in the second part (B). This resulted in significantly lower total OMNI scores in part B (95% CI [−3.87, −1.29], p < .001; Figure 2). Peak OMNI scores were seen at day 7 after HSCT, and the individual OMNI variables, mouth dryness and edema, contributed most to the total OMNI score each day (Figures 2 and 3). The highest individual OMNI scores for pain and lesions were found in part A (Figure 3).

| Salivary flow rate, pH and total protein
Both parts showed similar changes in flow rate reductions for UWS and SWS over the study period, with generally flow rate levels being lower in part A (Figure 4a and b); especially for SWS, both parts con-  Figure 6). The albumin levels in blood showed a decreasing trend over time with lowest levels between days 9 and 10 after HSCT ( Figure 7d).

| Correlations
Correlations between OMNI scores and salivary parameters are shown in Table 2. A positive correlation was found between total OMNI scores and SWS MUC5B, SWS albumin and SWS lactoferrin levels (R 2 = .318, p = .043; R 2 = .56, p = .029 and R 2 = .49, p = .043, respectively). In addition, total OMNI scores and UWS albumin levels were close to a statistical significant positive correlation (R 2 = .47, p = .05), and total OMNI scores and SWS flow rate were close to a statistical significant negative correlation (R 2 = −.39, p = .089).
As oral dryness is the largest contributor to OMNI scores in this study (Figure 2b), correlations were also determined between the dryness component of the OMNI score and flow rate and MUC5B levels in UWS and SWS (Table 2) F I G U R E 6 Mean ± 95% CI of total protein concentration in unstimulated whole-mouth saliva (UWS) (a) and stimulated whole-mouth saliva (SWS) (b) for part A (grey lines) and part B (black lines). Numbers in the graphs represent the number of patients per time point for part A (grey) and part B (black). Significance levels are based on paired t tests with scaled scores all compared to day −3, *p < 0.05, **p < 0.01, ***p < .001

(a) (b)
development, which resulted in significantly lower max OMNI scores during the whole study period in the second part with cryotherapy.
Consistent with the published literature, peak OM was reached around day 7 and had almost returned to baseline at the end of the study period (Blijlevens et al., 2008). Oral mucositis caused by HDM was reduced by the use of cryotherapy in this study. As melphalan is found in saliva, the local vasoconstriction caused by the cryotherapy may reduce the melphalan concentration in saliva (Epstein et al., 2002;Slavik, Wu, & Riley, 1993). Similar to a retrospective study, cryotherapy resulted in reduced OM, but did not reduce the duration of OM (Batlle et al., 2014).
Although a direct effect of HDM on salivary glands is unclear, conditioning therapy before HSCT causes salivary hypofunction (Epstein et al., 2002). Furthermore, salivary gland dysfunction by lymphocyte infiltration is found using gland biopsies and a scintigraphic method several weeks after HSCT with conditioning regimens consisting of cyclophosphamide, busulfan and TBI (Coracin et al., 2006;Lindahl, Lönnquist, & Hedfors, 1988). Most other studies measuring salivary flow rate after HSCT found a decrease in salivary flow rate (Amaral et al., 2012;Avivi et al., 2009;Boer, Correa, Tenuta, Souza, & Vigorito, 2015;Elad et al., 2006;Nagler, Barness-Hadar, Lieba, & Nagler, 2006). However, one study found no changes in UWS flow rate in autologous HSCT patients, while a study in allogeneic patients showed an increase in UWS flow rate 8-10 days after HSCT (Avivi et al., 2009;Boer et al., 2015). This increase might be related to dysphagia resulting in an increased resting volume of saliva as suggested by the authors (Boer et al., 2015).
One indication for glandular dysfunction caused by HDM is the similar decrease in flow rate and pH at day 4 in this study. However, the protein secretory functions of the glands are not affected as suggested by the increase in total protein and lactoferrin. Lactoferrin in saliva has a glandular origin although it is also a product of PMNs (Reitamo, Konttinen, & Segerberg-Konttinen, 1980). The increase in lactoferrin observed in this study may be derived from glandular origin as the PMN levels in blood decrease after HSCT resulting in neutropenia. To further investigate protein output of the salivary glands, salivary amylase levels could be studied or glandular saliva could be used to determine changes in specific glandular-derived proteins.
Myeloperoxidase is secreted by PMNs and could therefore be used as an indirect measurement of the PMN levels in saliva. The decrease in MPO found in this study at day 7 probably reflects neutropenia. Interestingly, the MPO levels in saliva showed an increasing trend some days before the PMNs were again detectable in blood.
This finding is supported by two other studies measuring PMN levels directly after HSCT and may indicate that PMNs from the stem cells directly reach the site of inflammation before they circulate at detectable levels in blood (Lieschke et al., 1992;Pink et al., 2009). Besides a decreasing trend in salivary flow rate, a decrease in salivary antimicrobial defence is also described in the literature mainly by a decrease in total or secretory IgA levels after HSCT (Avivi et al., 2009;Chaushu et al., 1994;Dens et al., 1996;Norhagen-Engstrom, Hammarstrom, Lonnqvist, Soder, & Smith, 1988). One mechanism to overcome this decreased antimicrobial defence might be an increased secretion of lactoferrin as most salivary proteins act in synergy for the antimicrobial function (Avivi et al., 2009;Imanguli et al., 2007). This hypothesis is supported in our study by a decreasing trend in total IgA levels and increased levels of lactoferrin. Furthermore, the increase in lactoferrin and albumin levels was also positively correlated with OM. Salivary albumin levels were increased during peak OM, while blood levels of albumin show a decreasing trend. This supports the previously described hypothesis of decreased integrity of the oral mucosa, which may result in leakage of blood substances into saliva during OM (Avivi et al., 2009;Izutsu et al., 1981).
Another OM-related complaint after HSCT is xerostomia, the feeling of a dry mouth. UWS and SWS. In resting conditions, saliva is mainly formed by secretions from the submandibular and sublingual glands. In stimulated saliva, the parotid gland contributes more, which results in more watery saliva compared to UWS (Proctor, 2016).
One limitation of several studies evaluating salivary changes in HSCT recipients is the mixed patient population. Only one earlier study evaluated early salivary changes in a homogenous group of MM patients receiving HDM and autologous HSCT (Avivi et al., 2009). Our observational study also included a homogenous patient population. However, both studies included small numbers of patients and lost power due to missing samples caused by nausea and hyposalivation especially during the OM period. Furthermore, due to logistic difficulties it was not possible to collect saliva at the same day during the whole study period. It was tried to collect saliva at the same time per patients, but also due to logistic difficulties, this was not always possible.
Salivary changes may influence the severity of OM. For proper symptom management, local salivary changes may be indicative.
This study found decreased total IgA levels in saliva resulting in decreased antimicrobial defence, which may be compensated by other antimicrobial proteins in saliva such as lactoferrin. However, further research is necessary to confirm and expand these results.
In conclusion, our observational study showed compositional changes in saliva reflecting inflammation of the salivary glands and the oral mucosae in the first days after autologous HSCT, of which the changes in albumin and lactoferrin were correlated with OM development. Although cryotherapy in the second study resulted in decreased OM severity, both studies showed similar trends in altered salivary protein composition.

ACK N OWLED G EM ENTS
The study was funded by the Radboudumc in Nijmegen (The Netherlands). For the optimization of the laboratory protocols to determine protein concentrations, we want to thank G.W. Hameete.

CO N FLI C T O F I NTE R E S T
None to declare.

AUTH O R CO NTR I B UTI O N
The study was designed by C.