Performance of Manufacturer Cleaning Recommendations Applied to 3D Food Ink Capsules for the Control of a Human Norovirus Surrogate

With the widespread availability of 3D food printing systems for purchase, users can customize their food in new ways. Manufacturer recommendations for cleaning these machines remain untested with regard to the prevention of foodborne pathogen transmission. This study aimed to determine if manufacturer cleaning recommendations for food ink capsules utilized in 3D food printers are adequate to control human norovirus (HuNoV). A HuNoV surrogate, Tulane virus (TuV; ~ 6 log10 PFU/mL), was inoculated onto the interior surface of stainless steel food ink capsules. Capsules were either unsoiled or soiled with one of the following: butter, protein powder solution, powdered sugar solution, or a mixture containing all three food components. The capsules were allowed to dry and then one of three hygienic protocols was applied: manual washing (MW), a dishwasher speed cycle (DSC), or a dishwasher heavy cycle (DHC). The interaction effect between DSC and pure butter was a significant predictor of log reduction (P = 0.0067), with the pure butter and DSC combination achieving an estimated mean log reduction of 4.83 (95% CI 4.13, 5.59). The DSC was the least effective method of cleaning when compared with MW and the DHC. The 3-way interaction effects between wash type, soil, and capsule position were a significant predictor of log reduction (P = 0.00341). Capsules with butter in the DSC achieved an estimated mean log reduction of 2.81 (95% CI 2.80, 2.83) for the front-most position versus 6.35 (95% CI 6.33, 6.37) for the back-most position. Soil matrix, cleaning protocol, and capsule position all significantly impact capsule cleanability and potential food safety risk. The DHC is recommended for all capsules, and the corners should be avoided when placing capsules into the dishwasher. The current study seeks to provide recommendations for users of additive manufacturing and 3D food printing including consumers, restaurants, industry, and regulatory industries.

AM technology poses sui generis foodborne illness risk due to increased handling demands: the food ink must be prepared via traditional handling methods, the capsules are then loaded with the desired food ink, and the capsule tag, capsule press, capsule holder are added before placing the assembled capsule into the 3D food printer (3DFP) for extrusion. Additionally, the food ink capsules can pose additional risks as they may contain niches in their nonplanar stainless steel surfaces that could become harborage sites for pathogens.

3
The global 3D food printing market is anticipated to reach USD$525.6 million by 2023 (46.1% annual increase), with the market shares divided into four main sectors: (1) commercial (43.5%), (2) government (25.8%), (3) hospital (20.8%), and (4) household (9.9%) (Lee, 2021). A comprehensive list of available 3DFPs and related patents has previously been published by Baiano (2020) and includes the 3DFP utilized in the current study, the Foodini (Natural Machines, Spain) as well as the first 3DFP, the Fab@Home (Creative Machines Lab, USA). Scallan and coauthors (2011) estimated the annual number of cases of human norovirus (HuNoV) in the USA was 5,461,731 with 149 associated deaths. Scharff (2012) estimated the total health-related cost associated with these cases was USD$3,677,000,000. Human norovirus can be transferred via direct person-to-person contact or ingestion of contaminated food or water or via indirect contact with fomites (Koopmans & Duizer, 2004;Mboko, et al., 2022). The largest source of contamination is human fecal matter which may contain 10 5 to 10 9 virions/g during a HuNoV infection (Koopmans & Duizer, 2004;Teunis et al., 2014), and the infective dose can be as low as 10-100 virions with an infective dose (ID 50 ) of 1015 genome copies or 18 viruses (Atmar & Estes, 2006;Hutson et al., 2004;Teunis et al., 2008;). Human fecal matter most commonly contaminates fresh produce and ready-to-eat (RTE) foods via workers' hands or contaminated water (Linscott, 2011). Given the extra handling demands of 3DFP, HuNoV is a pathogen of concern.
Studies of viral persistence on both food commodities and fomites have been completed. For example, Verhaelen et al. (2012) reported that HuNoV GII.4 and GI.4 RNA remained detectable after 1 week of storage at 4 °C, 10 °C, and 21 °C when inoculated on the surfaces of strawberries and raspberries. On stainless steel, ceramic, and Formica coupons, HuNoV underwent 1.50, 1.58, and 2.93 log 10 reductions, respectively, after 28 days of storage under ambient conditions (Liu et al., 2009). Notably, the studies by Liu et al. (2009) andVerhaelen et al. (2012) use RNA detection methods to quantify viral persistence. A limitation of relying on detection of RNA for persistence studies is that the infectivity of the remaining viruses is not known. Therefore, it is likely that RNA-based detection methods overestimate values of human norovirus persistence (Knight et al., 2016). Along with HuNoV persistence, enteric viruses can be very difficult to eradicate using common food treatments intended to control bacteria such as pH, temperature, irradiation, and water activity (Hirneisen et al., 2010).
Cultivation of HuNoVs is still very expensive and difficult (Estes et al., 2019) and thus surrogates are often used. Cromeans and coauthors (2014) investigated the responses of HuNoV, Tulane virus (TuV), murine norovirus (MNV), feline calicivirus (FCV), Aichi virus, and porcine enteric calicivirus after exposure to heating at 56 °C, alcohols, chlorine on surfaces, extreme pH, and high hydrostatic pressure. Tulane virus was found to be one of the most resistant surrogates to all treatments except high hydrostatic pressure. In addition, TuV is in the Caliciviridae family and binds to histo-blood group antigens similar to HuNoV genotypes of public health significance (i.e., GII.4) (Farkas et al., 2010). The environmental resistance, cultivatibility, and similar physicochemical properties of TuV to HuNoV lead to its selection for this study.
Overall, there is a paucity of research into the risks of foodborne illness associated with food extruded from 3DFPs. Safety data related to 3D printed foods and 3DFPs is essential not only for regulatory action aimed at mitigation of risks but also consumer confidence in the foods being produced. Therefore, the present study focuses on the efficacy of cleaning practices recommended by manufacturers to remove or inactivate TuV from contaminated food ink capsules. Specifically, the study objectives were to (i) evaluate manufacturer cleaning recommendations for the removal of TuV from food ink capsules; (ii) investigate how soil composition and presence impact food ink capsule cleanability; and (iii) determine how food ink capsule placement in the dishwasher impacts food ink capsule cleanability.

Tulane Virus Propagation
Tulane virus, provided by Dr. Jason Jiang (Cincinnati Children's Hospital Medical Center, Cincinnati, OH) was propagated in LLC-MK2 cells (ATCC CCL-7) (American Type Culture Collection, Manassas, VA) as previously described with slight modifications (Arthur & Gibson, 2015). Briefly, LLC-MK2 cells were grown in complete growth media [Medium 199 with Earle's Balanced Salt Solution (EBSS) (Hyclone, Logan, UT) supplemented with 1% Penicillin/ Streptomycin (Hyclone), 1% Amphotericin B (Corning, Mediatech Inc., Manassas, VA), and 10% Fetal Bovine Serum (FBS) (Hyclone)]. Cell monolayers with 90% LLC-MK2 cell confluency were inoculated with TuV stock at a multiplicity of infection (MOI) of 0.1 in 6 mL Opti-MEM (Gibco Life Technologies, Grand Island, NY) and incubated with continuous rocking at 5% CO 2 and 37 °C. After 1 h, 20 mL of Opti-MEM supplemented with 2% FBS was added to the T150 flask. The flask was incubated without rocking at 5% CO 2 and 37 °C until a complete cytopathic effect (CPE) was observed (approximately 60 h). Once CPE was observed, the flask was transferred to − 80 °C and held overnight. The following day, TuV was harvested via three freeze-thaw cycles and centrifugation at 3000×g for 15 min at 4 °C. The supernatant was filtered through a Steriflip® 1 3 0.45 µm filter (Millipore, Burlington, MA) and then aliquoted in 1 mL cryovials before storing at − 80 °C.

Quantification of TuV
To enumerate TuV, six-well tissue culture-treated plates were seeded with 4 × 10 5 LLC-MK2 cells per well and incubated for 24 h at 37 °C under 5% CO 2 . The media was aspirated before the cells were inoculated with 500 μL of tenfold serially diluted TuV stock (10 -1 to 10 -6 ) in Opti-MEM supplemented with 2% FBS. The plates were incubated at 37 °C under 5% CO 2 with gentle rocking for 1 h. The overlay mixture was prepared by melting 3% low-melting NuSieve™ GTG™ Agarose (Lonza, Walkersville, MD) and combining it with an equal volume of Opti-MEM supplemented with 2% FBS. After incubation, the sample was aspirated, and 2 mL of the overlay mixture were slowly added to each well. After the overlay solidified (approximately 15 min), the six-well plates were incubated at 37 °C and 5% CO 2 until plaques appeared (approximately 96-120 h). To visualize the plaques, 3% of neutral red stock (Sigma-Aldrich, St Louis, USA) was prepared in 1X PBS, and 2 mL were added to each well before incubation at 37 °C under 5% CO 2 for 3 h. After incubation, the stain was aspirated, and the plates were examined for plaque-forming units (PFU). Tulane virus stock concentrations were approximately 10 6 PFU/mL.

Food Matrix Preparation
To prepare the complex soil matrix unsalted butter (Member's Mark Unsalted Sweet Cream Butter; Sam's West, Inc, Midwest City, OK), protein powder (Sports Research Collagen Powder Supplement; Sports Research, San Pedro, CA), and powdered sugar (Domino Confectioners Sugar; Domino Food, Yonkers, NY) were procured and combined to create a 1:1:1 matrix of fat, protein, and carbohydrate (25 g butter, 20 g powdered sugar, 21 g protein powder, and 50 mL deionized [DI] water). Individual food components were also tested separately and used in pure form (butter) or mixed with DI water to form aqueous preparations (protein powder and powdered sugar) suitable for printing by the 3DFP.

Preparation and Inoculation of the Food Ink Capsules
The inoculation method was adapted from Moore and Griffith (2002) and Keeratipibul et al. (2017). The general shape of a Foodini food ink capsule can be approximated by a cylinder with a length of 11 cm and a diameter of 4 cm. The capsule nozzle has a length of 2.5 cm, and a diameter of 4 cm, including the tip from which food ink is extruded (Fig. 1). The Foodini user manual indicates that each capsule should contain no more than 100 mL of food ink.
All stainless steel capsules were washed with dish soap (Dawn Ultra; Proctor and Gamble, Cincinnati, OH) before being rinsed in DI water and sprayed with 70% ethanol. After being allowed to air dry, the capsules were autoclaved for 15 min at 121 °C in sterilization pouches.
The capsules were then inoculated with 250 μL of 10 6 PFU/mL TuV stock that was distributed evenly with a sterile inoculation loop. After being allowed to dry in a biosafety cabinet for 30 min, 10 g of soil material (see Food Matrix Preparation) was added and distributed evenly with an inoculation loop, if applicable. Then, the capsules were subjected to either a cleaning protocol or control procedure (Fig. 2).

Cleaning Schema
Manufacturer recommendations for cleaning food ink capsules are vague with both dishwashing and manual washing being recommended. However, no further specifics on dishwasher setting, water temperature, or washing times are provided. The three cleaning schema were chosen to represent the strongest dishwasher setting, the weakest dishwasher setting, and what the authors believe most closely resembles manual washing in the homes of consumers.
In the manual washing (MW) protocol, dishwashing soap (Dawn Ultra; Proctor and Gamble) was applied to a dishwashing sponge, and the capsules were individually washed under water (50 ± 2 °C) for 60 s. In the dishwashing protocols, a portable countertop dishwasher (SPT SD-2225DWA Energy Star Countertop Dishwasher; Sunpentown International, City of Industry, CA) was utilized under two different cycles ["heavy" (DHC) and "speed" (DSC)]. The DHC is a 2.33-h cycle with a maximum temperature of 67.7 °C, while the DSC is a 0.75-h cycle with a maximum temperature of 55.0 °C. Both cycles used the same amount (18 g) of dishwashing detergent gel (Cascade Complete; Proctor and Gamble). The capsules were placed in the right-most and front-most three positions within the dishwasher (Fig. 2) as this was the most intuitive and unobstructed location due to the layout of the dishwashing rack and the length and angle of the support protrusions.

Sampling of the Food Ink Capsules
A polyurethane foam (PUF) swab (PUR-Blue™ Dry Swab; World Bioproducts, Woodinville, Washington) was premoistened with 3 mL of 1× PBS and used to swab the capsule interior for 60 s. The swab was returned to the same 3 mL of 1 PBS before being vortexed for 30 s. The sample was transferred to a 2 mL microcentrifuge tube and serially diluted in Opti-MEM supplemented with 2% FBS, 1% Penicillin/ Streptomycin, and 1% Amphotericin B. As negative control, PBS was tested alongside a sterile PUF swab on a sterile capsule. For method evaluation and positive controls, a capsule was inoculated with TuV (and soiled, if required), allowed to dry, and swabbed. The method was validated for recovery in duplicate experiments with triplicate replication with an average recovery efficiency of ~ 84%. The samples were enumerated using virus plaque assay (see Quantification of TuV).

Statistical Analysis
Two experimental trials with three replicates each were performed per cleaning protocol and soil matrix combination. Log reductions were calculated as −log 10 FinalPFU InitialPFU . Data were analyzed in R Studio using a GLM with quasipoisson errors to account for the observed overdispersion. The treatment means and their associated standard errors were calculated using estimated marginal means (emmeans). Statistical differences between treatments were determined using multiple pairwise comparisons and visualized using compact letter display.
A second analysis was completed to determine if soiling material, capsule position, and dishwasher cycle were significant predictors of log reduction in TuV. Data were modeled in R Studio using ANOVA as the QQ plot didn't stray far from normality and the analysis of variance showed homoscedasticity. The treatment means and their associated 95% confidence intervals were calculated using estimated marginal means (emmeans). Statistical differences between treatments were determined using multiple pairwise comparisons and visualized using compact letter display.

The Effect of Cleaning Protocol and Soil Matrix on Viral Reduction
All raw data are plotted in Fig. 3, which shows measured log reduction against the initial PFU per capsule based on the cleaning protocol and the soil composition. Note that reductions are estimated when results were below the limit of detection (0.78 log 10 PFU/mL). The 2-way interaction effect between DSC and pure butter was a significant predictor of estimated mean log reduction (P = 0.0067), so conclusions about the main effects cannot be made. The estimated mean log reductions alongside their 95% confidence intervals and the statistical grouping from the post-hoc analysis are shown in Fig. 4, where it can be observed that the DSC performed least well at removing or inactivating TuV. This observation is especially apparent when combined with the pure butter soil. Manual washing was not statistically different from any other treatment, and the DHC consistently led to TuV removal below the limit of detection. Figure 5 supplies the raw data from the dishwasher placement comparisons. The 3-way interaction effect between dishwashing cycle, capsule position, and soil matrix was a significant predictor of log reduction (P = 0.00341), so conclusions cannot be made about 2-way interactions or main effects. As shown in Fig. 6, position 1 (the front-most position) in conjunction with the DSC was statistically different from all other positions during either wash cycle. Differences between food matrices are also elucidated by this analysis, where the sugar solution was statistically different from both the complex matrix and butter, but not from the Fig. 3 Raw data of log 10 reductions based on cleaning protocol (color) and soil type (shape). Many of the log reductions were below the limit of detection (0.78 log 10 PFU/mL) and were plotted as completely reduced. DHC dishwasher heavy cycle, DSC dishwasher speed cycle, MW manual washing Fig. 4 Generalized Linear Model of estimated mean log 10 reductions with quasipoisson error approximations based on cleaning protocol and soil type. Error bars represent 95% confidence intervals, and compact letter format is used to designate statistical differences between treatments at P = 0.05. DHC dishwasher heavy cycle, DSC dishwasher speed cycle, MW manual washing Fig. 5 Raw data of log reductions based on dishwasher cycle (shape) and capsule position (color). Capsule position 1 represents the capsules loaded into the front-most corners (closest to the door), capsule position 2 indicates the capsules directly behind capsule 1, and capsule position 3 indicates the capsules directly behind capsule position 2. Many of the log reductions were below the limit of detection (0.78 log 10 PFU/mL) and were plotted as completely reduced 1 3 protein solution (Fig. 6). Also, positions 2 and 3 were not statistically different between the DSC and DHC. These data demonstrate that capsule position was likely a confounding variable in our first analysis, and this second analysis is necessary in order to see differences between the two dishwasher cycles.

Discussion
Human norovirus outbreaks are common in crowded locations such as cruise ships (Wikswo et al., 2011), nursing homes (Calderon-Margalit, 2004, schools (Yu et al., 2011), and even hospitals (Johnston et al., 2007). While the report written by Wilkswo et al. (2011) was not able to directly link poor dishwashing conditions with the HuNoV outbreak, the authors did note that a dishwasher utilized to clean buffet dishes was operating below the required minimum temperature during the sanitizing rinse, and that items soiled with food debris (due to inadequate washing) were being stored with clean dishes and used to service the buffet. Crowded places such as those listed, are prime locations for the implementation of 3DFP processes. Therefore, having reliable recommendations for the cleaning of 3DFP food ink capsules will be important in preventing or mitigating future HuNoV outbreaks.
In the present study, manufacturer cleaning recommendations for the removal/inactivation of TuV as a surrogate for HuNoV were evaluated. When comparing the efficacy of the three washing protocols examined (DSC, DHC, MW), only the DSC in combination with the pure butter soil was found to be statistically unique at P = 0.05, and the interaction between the DSC and pure butter was found to be a significant predictor of estimated mean log reduction (P = 0.0067).
In Figs. 3 and 4, it can be observed that the DHC always performed the best at removing/inactivating TuV (complete log reductions were always measured). Manual washing procedures performed second-best (with the exception of the sugar solution soil), but it was not statistically different from the DHC in any instance. The DSC performed the worst (with the exception of the sugar solution soil), but it was not significantly different from MW in any instance and was only significantly different from DHC when pure butter soil was utilized. This general pattern of efficacy was also found in a similar study by the authors that examined Salmonella Typhimurium and Listeria monocytogenes removal from soiled 3DFP capsules (unpublished data). However, in the unpublished study, it was the complex soil mixture, not pure butter, that had a significant interaction effect with the DSC. This may indicate differences between bacteria and viruses insofar as the type of shielding provided by soils with different macromolecule compositions.
To the authors' knowledge, no studies have explicitly examined the potential relationship between the macromolecule composition of a food and thermal protection in HuNoVs. However, in a study of Hepatitis A virus (HAV), Deboosere and coauthors (2004) varied the sucrose concentration in strawberry mashes at 85 °C and found D-values of 0.96, 2.37, and 4.98 min at 28°Brix, 40°Brix, and 52°Brix, respectively. Bidawid and coauthors (2000) conducted a similar experiment with HAV to examine the relationship of milkfat content on thermal inactivation. Those authors reported a time for each of the first five log 10 reductions for three different fat concentrations (1% fat skim milk, 3.5% fat homogenized milk, and 18% cream) and eight different temperatures (65 °C, 67 °C, 69 °C, 71 °C, 73 °C, 75 °C, 80 °C, 85 °C). Overall, the protective nature of increased fat content was demonstrated in every condition (except at Fig. 6 Analysis of variance with multiple pairwise comparisons of estimated mean log reductions based on dishwasher cycle and capsule position in the dishwasher. Capsule position 1 represents the capsules loaded into the front-most corners (closest to the door), capsule position 2 indicates the capsules directly behind capsule 1, and capsule position 3 indicates the capsules directly behind capsule position 2. Error bars represent 95% confidence intervals, and compact letter format is used to designate statistical differences between treatments at P = 0.05. DHC dishwasher heavy cycle, DSC dishwasher speed cycle 1 3 85 °C) (Bidawid et al., 2000). The thermal inactivation of poliovirus in ground beef at three different fat contents (3% fat, 27% fat, and 47% fat) and four different temperatures (50 °C, 60 °C, 70 °C, and 80 °C), demonstrated the protective nature of fat in meat. Unfortunately, the authors did not investigate the role of protein content (Filppi & Banwart, 1974). To the authors' knowledge, no studies examining the protective role of protein in the thermal inactivation of viruses have been published.
While poliovirus and HAV are less than ideal surrogates for HuNoV, the data provided in the studies by Deboosere et al. (2004), Bidawid et al. (2000), and Filppi and Banwart (1974) can likely be used to show general patterns of protection as the macromolecule concentration increases, but not estimated D-values or log 10 reductions. Also, only Filppi and Banwart (1974) studied temperature ranges relevant to the DSC. Studies on the protective nature of fat, carbohydrate, and protein during thermal treatment of viruses under various dishwashing conditions should be completed with suitable HuNoV surrogates such as MNV or TuV. In the present experiment, it is not possible to conclude whether TuV is being removed or thermally/chemically inactivated, but future studies could be completed using methods that assess viral capsid integrity such as the use of intercalating dyes combined with polymerase chain reaction (Manual et al., 2018). When considering capsule position during DSC conditions (55 °C for 45 min) (Fig. 6), the protectiveness of the matrices ranked from best to worst are (1) pure butter, (2) complex soil matrix, (3) protein powder solution, and (4) sugar solution. While no authors have compared the protective nature of different macromolecules with one another, there is evidence that fat and sugar can play a protective role when viruses are exposed to challenging thermal conditions (Bidawid et al., 2000;Deboosere et al., 2004;Filppi & Banwart, 1974). In addition, the present study also supports that protein may play a similar protective role. Lucassen et al. (2021) investigated MNV as a surrogate for HuNoV and found that a combination of a bleach-containing dishwasher detergent, a cleaning temperature of 45 °C for 45 min, and a rinsing temperature of 50 °C led to 4.2 log 10 reductions; however, only limited conditions were tested (i.e., other dishwasher cycle times and manual washing were not tested). Rinsing temperature was tested at 30 °C, 50 °C, and 70 °C, and in agreement with the results of the present study, increased temperature led to a significant increase in viral log reduction (Lucassen et al., 2021).
To the authors' knowledge, no prior studies have examined washing efficacy with TuV as a HuNoV surrogate. Feliciano et al. (2012) conducted a study with MNV to examine the sanitization efficacy of manual and mechanical washing protocols used in restaurants. The authors used ceramic plates, drinking glasses, and stainless steel forks as food contact surfaces and contaminated them with cream cheese or reduced-fat milk soils. The mean reductions of MNV on the plates, forks, and drinking glasses after the washing treatment without sanitization were, respectively, 2.6, 1.3, and 0.7 log 10 PFU/mL for mechanical washing and 2.8, 1.1, and 1 log 10 PFU/mL for manual washing. The mean reductions achieved after mechanical washing with chlorine sanitization and washing with quaternary ammonium (QAC) sanitization were not statistically different (P > 0.0001) from those achieved by the control for either manual or mechanical washing. Fortunately, the viral counts detected on the different surfaces after washing and sanitization were statistically different (P < 0.0001) from the initial viral counts prior to the washing, indicating that washing does have a statistically positive effect, while sanitization had none. Unfortunately, the authors did not directly compare manual washing with mechanical washing (Feliciano et al., 2012).
Individual capsule placement was investigated based on within-group variability. Moreover, even visual inspection suggested that the first capsule position (Fig. 2) was being cleaned to a lesser extent than other positions during the DSC (Figs. 5, 6). All positions were cleaned equally (P > 0.05) in the DHC, and positions 2 and 3 in the DSC were not statistically different from any position in the DHC. Moreover, unsoiled capsules were not statistically different from the prior two groupings. When capsules were soiled, TuV in capsule position 1 underwent fewer log reductions than in any other position in the DSC at P = 0.05. This difference was especially pronounced for the complex soil mixture and pure butter, which were both significantly different from the sugar solution, but not from the protein solution. All soiled conditions in capsule position 1 were significantly different from the unsoiled conditions, DHC, and positions 2 and 3 in the DSC. The observed inconsistencies in cleanliness across the capsule positions are thought to be caused due to the dimensions and construction of the dishwasher. Briefly, the authors have observed that the dishwasher arm is not able to fully access the corners of the dishwasher due to geometric limitations, resulting in inconsistent cleaning. No previous studies on dishwasher placement and pathogen reduction are known to the authors.
Based on these data, the authors continue to recommend that 3DFP capsules be washed on the highest available dishwasher setting and suggest that capsules be placed away from the corners where the dishwasher arm may be less effective. This research seeks to continue to develop the field of AM hygiene as it relates to foodborne illness risk and hygienic recommendations. Additionally, the authors seek to develop the foundation of understanding hygienic efficacy with regard to foodborne pathogen reduction in dishwashers and by manual washing of 3DFP food ink capsules. These results elucidate weaknesses in manufacturer recommendations for cleaning of 3DFP food ink capsules that should be addressed in the manuals which are distributed with the machines.
In future, this study should be repeated using other foodborne viruses, HuNoV surrogates, and HuNoV strains when their cultivation becomes more feasible. More ratios and sources of macronutrients should be explored. For example, pea protein powder and cricket protein powder may yield different results from the beef-based protein used in the present study. More dishwasher settings, detergents, and models should be tested, as should the presence of other soiled objects in the dishwasher. Different MW conditions should be tested, such as colder water and shorter washing times. Finally, contamination location on the food ink capsules should be varied. For example, the threads and rubber O-rings present niches not available in the main body of the stainless steel capsule.

Conclusions
In conclusion, manufacturer recommendations may be inadequate for preventing foodborne illness caused by HuNoV. The findings of the present study indicate lipid-based soils may present the largest challenge to hygienic efforts. Moreover, the longest and hottest dishwasher cycle available should always be utilized, and the capsule position in the dishwasher should not be localized in the corners where the dishwasher arm may struggle to clean effectively.
Author Contributions ANH and KEG contributed to conceptualization, methodology, and writing-reviewing and editing; ANH contributed to formal analysis, investigation, visualization, and writing-original draft; and KEG contributed to resources, supervision, and funding acquisition.
Funding This work is supported in part by the National Institute of Food and Agriculture (NIFA), US Department of Agriculture (USDA) Hatch Act funding, and the University of Arkansas Distinguished Doctoral Fellowship (DDF). K.E. Gibson also received a Faculty Equipment and Technology grant from the University of Arkansas Honors College used to support the purchase of the 3D food printer used in the present study.

Data Availability
The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.

Declarations
Competing interests No conflict of interest declared.