The impact of thermal pasteurization on viral load in human milk and other matrices: A rapid review

Holder pasteurization (62.5ºC, 30 min) of human milk (HM) is thought to reduce the risk of transmitting viruses to an infant. Some viruses may be secreted into milk – others may be contaminants. The effect of thermal pasteurization on viruses in HM has yet to be rigorously reviewed. The objective of this study is to characterize the effect of commonly used pasteurization techniques on viruses in HM and non-HM matrices. Databases (MEDLINE, Embase, Web of Science) were searched from inception to April 20 th , 2020 for primary research articles assessing the impact of pasteurization on viral load or detection of live virus. Reviews were excluded, as were studies lacking quantitative measurements or those assessing pasteurization as a component of a larger process. Overall, 65,131 reports were identified, and 108 studies included. Pasteurization of HM at a minimum temperature of 56ºC-60ºC is effective at reducing detectable live virus. In cell culture media or plasma, coronaviruses (e.g., SARS-CoV, SARS-CoV-2, MERS) are highly susceptible to heating at ≥ 56ºC. Although pasteurization parameters and matrices reported vary, all viruses studied, with the exception of parvoviruses, were susceptible to thermal killing. Future research important for the study of novel viruses


Introduction
Since the emergence of SARS-CoV-2 in late 2019, ensuring that current high-quality screening, handling and pasteurization standards are sufficient for maintaining a safe supply of human donor milk has been an ongoing challenge for milk banks (Furlow 2020).Human donor milk is used as a bridge for hospitalized infants while their mother's own milk supply is being established; among very low birth weight infants, the use of human donor milk instead of preterm formula as a bridge has been shown to reduce the incidence of necrotizing enterocolitis (Underwood 2013;Quigley et al. 2019).Milk banking associations, including the Human Milk Banking Association of North America and the European Milk Banking Association have responded to the pandemic by issuing new guidelines with respect to enhanced donor screening, including asking specific questions to assess the likelihood of a potential donor being infected with SARS-CoV-2("COVID-19: EMBA Position Statement" 2020; "Milk Banking and COVID-19" 2020).While all donor milk from non-profit milk banks in North America is pasteurized using the Holder method (62.5ºC, 30 min) to inactivate potentially pathogenic bacteria and viruses, additional research is warranted to determine whether SARS-CoV-2, is inactivated by Holder pasteurization (Arslanoglu et al. 2010; Guidelines for the Establishment and Operation of a Donor Human Milk Bank 2018).
At present, the virome of human milk has been understudied.Few studies have investigated whether or not viruses that may cause disease in preterm infants are present in human milk (Mohandas and Pannaraj 2020).Viruses may be present in human milk as a result of secretion into the milk in the mammary tissue, notably, cytomegalovirus, human t-lymphocytic virus, and human immunodeficiency virus (HIV), or may be present as a contaminant from skin or respiratory droplets either in the milk or on the containers (Michie 2001).Regardless of origin, it is important to understand how viruses found in human milk respond to thermal pasteurization.
To date, there has been no systematic review of the impact of thermal pasteurization on viral load or live virus detection in a human milk matrix or other non-human milk matrices.The primary aim of this review is to characterize studies conducted in human milk to determine how certain viral families that are either present in human milk, or used as surrogates, respond to thermal pasteurization as assessed by viral load or live virus detection.To expand the scope of viruses tested, the secondary objective is to summarize studies conducted in non-human milk matrices that have examined the effect of thermal pasteurization on any virus.This review also aims to compare viruses that have been assessed in studies using both human milk and nonhuman milk matrices to ascertain any trends in susceptibility to thermal pasteurization.

Materials and methods
The Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines were followed in completion of this rapid review, except where indicated (Moher et al. 2009).This rapid review is in response to the COVID-19 pandemic.

Search strategy and selection criteria
References for this rapid review were identified through electronic searches of various online databases including MEDLINE, Embase and Web of Science, from database inception to April 20 th , 2020, with the assistance of a research librarian.The search strategy focused on keywords to identify articles that assessed the effect of thermal pasteurization or heat inactivation, including Holder pasteurization, on the detection of live virus or viral load in human milk or other non-human milk matrices.The names of viral families, as per the current taxonomic classification, were included in the search as they may be present in human milk (by secretion or .contamination) or could be used as surrogate viruses to model highly pathogenic or nonculturable viruses (King et al. 2012).
The keywords and MeSH terms included for all database searches were intended to capture all relevant research with respect to thermal pasteurization of viruses in human milk, the primary outcome of this rapid review.To increase the scope, we supplemented the search to capture research articles that tested all matrices other than human milk.The search strategy is summarized in Table S1 and included three main ideas.The first concept included viral taxonomic families using keywords and MeSH terms based on the nomenclature suggested by the International Committee on Taxonomy (King et al. 2012).The second concept consisted of synonyms and phrases closely related to human milk (e.g.breast milk, donor milk etc.).Lastly, the final concept was thermal pasteurization and its synonyms (e.g., Holder pasteurization, heat etc.).Our initial search aimed to retrieve articles specific to human milk which was achieved by combining all three concepts; by only retaining the first and last concept, a second set of articles was retrieved that theoretically involved thermal pasteurization and viruses in all other matrices, including human milk.Grey literature was searched as per the previously published guidelines including from dissertations and google advanced search (Natal 2019).Articles resulting from those searches and relevant references cited in those articles were reviewed.
After duplicates were removed, titles, then abstracts were screened by a single reviewer.Primary research articles were included if they assessed the effect of Holder pasteurization (62.5°C -63°C) or any other heat treatment on viral load or detection of live virus in human milk or other matrices.Eligible study designs included pre-post or longitudinal; in either design, the outcome, detection of live virus or viral load, was assessed before and after pasteurization, or at discrete time points during a given pasteurization process.Qualitative, observational and review studies were excluded, in addition to experimental studies that did not assess viral load (quantitative) or detectable live virus.Studies that investigated how thermal pasteurization and the addition of matrix stabilizers, affects viral load or live virus detection were also excluded; the outcome of these studies may be confounded by the fact that the integrity of viruses may be different as certain stabilizers are added or removed.Studies were also excluded if thermal pasteurization was tested in combination with other processing techniques, (e.g., irradiation, lyophilization during the production of plasma concentrates), unless the study was appropriately controlled.
The primary rationale being that aspects of processing, other than heat, may also lead to the inactivation of viruses.Reports on clinical trials or studies published in non-scientific journals were not included.All studies irrespective of language or year published were included.
Multiple attempts were made to retrieve the full-text of all articles screened on the basis of title and abstract including interlibrary loan and/or author follow-up.Data were extracted from eligible full-text articles including viruses tested, matrix used, thermal pasteurization parameters (temperature, time) and a measure of reduction in viral load/detectable virus.Included studies were summarized after being dichotomized into two groups depending on whether detectable live virus or viral load was tested in human milk or another matrix.To determine whether a human milk matrix affected the results, a subanalysis was conducted on studies that tested the same viruses in both human milk and non-human milk matrices.In this subanalysis, only studies that assessed virus presences by plaque reduction assay or endpoint dilution (TCID 50 ) were included.First, viruses that were tested in both groups were determined by cross-referencing; relevant data (log-reduction, temperature, and duration of pasteurization) was then extracted and aggregated.Unless otherwise defined, complete inactivation is a concentration of virus that was below the lower detection limit of the assay.If multiple studies assessed the same virus, the pasteurization conditions used in the summary were matched as closely as possible to the data available in studies experimenting with human milk.

Study selection and characteristics
The selection of studies is summarized in Figure 1.A total of n=65,131 reports were identified and assessed for eligibility.This included 23,441 citations from MEDLINE, 34,479 citations from Embase, 7,200 records from Web of Science and 11 from manual searches.Altogether, n=64,950 records were excluded on the basis of title and abstracts alone, encompassing articles that did not meet the inclusion criteria (n=44,287 records) or duplicate records (n=20,663).After title and abstract screening, 181 reports remained for full-text review.Upon full-text review, 73 reports were excluded: 6 were duplicate records, 2 could not be retrieved and 65 did not meet inclusion criteria.Thus, 108 articles were included in the review and were organized according to the matrix used in testing the effect of pasteurization on viral load.

Studies conducted in human milk
First, we summarized 18 unique studies that used human milk as the matrix to test the effect of pasteurization on thirteen different viruses (Table 1).Most studies reported on viral addition experiments, while few studies subjected milk with endogenous virus to thermal pasteurization.
Predominantly, the viruses tested were caspid enveloped and belonged to 8 different viral families including: caliciviridae, filoviridae, flaviviridae, herpesviridae, papillomaviridae, picornaviridae, retroviridae, and togaviridae.Cytomegalovirus and HIV were the most common viruses studied with 8 and 7 articles respectively.To assess surviving virus concentration following pasteurization, plaque reduction assays and endpoint titration assays (TCID 50 ) were most frequently used, although some studies used immunofluorescence, reverse-transcriptase enzymatic assays and secreted embryonic alkaline phosphatase (SEAP) reporter assay.
Based on the literature reviewed, Holder pasteurization, defined as a temperature of 62.5ºC -63ºC held for 30 minutes, resulted in complete inactivation of viruses in the herpesviridae family, including cytomegalovirus (Dworsky et al. 1982;Hamprecht et al. 2004;Donalisio et al. 2014); however, complete inactivation of herpes simplex virus did not occur, requiring a temperature of 100ºC, 5 min (Welsh et al. 1979).In fact, for cytomegalovirus specifically, some studies reported complete inactivation at 60ºC-63ºC for 5 seconds to 30 minutes (Friis and Andersen 1982;Klotz et al. 2018;Maschmann et al. 2019).Similarly, retroviridae were susceptible to heating in a human milk matrix whereby complete inactivation was observed after pasteurization above 60ºC, for a minimum of 5 seconds.In particular, flash heating and Holder pasteurization completely inactivated HIV-1 in human milk (Orloff et al. 1993;Israel-Ballard et al. 2007;Volk et al. 2010;Hoque et al. 2013); high temperature short time (72ºC for 8 seconds) similarly yielded complete inactivation (>5.5-log reduction) (Terpstra et al. 2007).Holder pasteurization was found to inactivate (>5-log reduction) Ebola virus and Marburg virus of the filoviridae family, Zika virus (>6-log reduction) of the flaviviridae family, Semliki forest virus of the togaviridae family (4.2-log reduction) and human papillomavirus of the papillomaviridae family (Welsh et al. 1979;Hamilton Spence et al. 2017;Pfaender et al. 2017).Some nonenveloped members of the picornaviridae family were found to be more resistant to heating (Terpstra et al. 2007); flash heating (72 ºC for 16 seconds) of hepatitis A virus and porcine parovirus yielded a 2-or 0.5-log reduction in TCID 50 /mL respectively.Infectivity of coxsackievirus persisted after Holder pasteurization, although reduced by 3.6-log PFU/mL (Welsh et al. 1979).

Studies conducted in non-human milk matrices
Second, we summarized the remaining 90 unique studies that were identified during the literature review that assessed the effect of thermal pasteurization on viruses in a non-human milk matrix (Table 2).Cell culture media was the most prevalent matrix used in testing; other common matrices included bovine milk, bovine serum, human serum albumin, human plasma.In total, 21 unique families of viruses were tested including: adenoviridae, anelloviridae, birnaviridae, caliciviridae, circonviridae, coronaviridae, flaviviridae, hepadnaviridae, hepeviridae, herpesviridae, orthomyxoviridae, paramyxoviridae, parvoviridae, picornaviridae, polymaviridae, poxviridae, reoviridae, retroviridae, rhabdoviridae, togaviridae.The majority of studies tested non-enveloped viruses in the families of picornaviridae (n=38), and caliciviridae (n=24), in addition to retroviridae (n=16).
Hepatitis A was the most commonly tested virus tested of the picornaviridae family and was seen to be particularly heat sensitive in a variety of matrices including bovine milk, cell culture media and soft-shell clams.For example, a minimum of a 4-log reduction in infectivity of Hepatitis A was observed after different thermal pasteurization parameters such as 60ºC-65ºC for 10-180 min (Croci et al. 1999;Bidawid et al. 2000;Gibson and Schwab 2011); to 72ºC for 1-13 min (Bidawid et al. 2000;Araud et al. 2016), and to 90ºC for 5 min (Sow et al. 2011).Murine norovirus, the most frequently tested virus of the caliciviridae family, was also observed to be sensitive to heat.A reduction in infectivity of greater than 5-log was observed at temperatures of 60ºC -67ºC for 1-60 min (Gibson and Schwab 2011;Shao et al. 2018), >3.5-log reduction at 72ºC for 1 min (Hewitt et al. 2009;Araud et al. 2016), and >5-log reduction at 85ºC -90ºC for 1-5 min (Sow et al. 2011;Park et al. 2014a).HIV-1 was the most commonly tested of the retroviridae and was also susceptible to heat-inactivation.Greater than 4-log reduction in TCID 50 was observed at 60ºC-65ºC for 10-15 min (Lelie et al. 1987;Gregersen et al. 1989) ; similar reductions were observed at 77ºC-80ºC after 0.25 seconds (Charm et al. 1992).

Viruses tested in human milk and other matrices
Finally, the summary of the comparisons among viruses that were tested in both a human milk and a non-human milk matrix is shown in Table 3. Overall, the range of temperatures that yielded some degree of log reduction were consistent among viruses, irrespective of the matrix.
Cytomegalovirus, for example, was a virus where there was good agreement among studies testing thermal pasteurization in either a human milk or a non-human milk matrix; inactivation was evident at temperatures between 50ºC and 65ºC for 10-30 min.Similarly consistent, porcine parvovirus in the parvoviridae family was found to be heat resistant in either human milk or nonhuman milk matrices (Danner et al. 1999;Terpstra et al. 2007;Sauerbrei and Wutzler 2009).
There were some differences in the time required for the log reduction in infectivity depending on matrix, but there were no discernable trends identified.

Discussion
Pasteurization is an essential part of human donor milk banking and is practiced worldwide to reduce or eliminate the risk of transmission of viruses that may be expressed in milk or may be found as a contaminant; Holder pasteurization (62.5ºC, 30 min) is the most common method used(Guidelines for the Establishment and Operation of a Donor Human Milk Bank 2018).Our rapid review aimed to summarize the literature pertaining to the effect of thermal pasteurization on viral load and detectable live virus; in particular, research that has been conducted using a human milk matrix.Our rapid review also aimed to compare viruses that have been both tested in a human milk matrix and a non-human milk matrix to better understand any potential modulating effects.
As expected, the most commonly studied viruses in human milk in relation to thermal pasteurization included those that have been previously shown to be transmitted through breastmilk; primarily cytomegalovirus and HIV-1 which are enveloped viruses belonging to the herpesviridae and retroviridae families respectively (Prendergast et al. 2019).Although not as common as cytomegalovirus or HIV, Ebola, Marburg, and Zika viruses have also been studied in human milk given that viral nucleic acid has been detected in milk and transmission is a potential concern (Hamilton Spence et al. 2017;Sampieri and Montero 2019).Despite differences in viral taxonomy and caspid envelope, pasteurization is effective at significantly reducing detectable virus or viral load by several log, and in many cases, to below detectable levels (Table 1).
Many studies involving human milk tested pasteurization parameters that included the Holder method (62.5°C, 30 minutes) in order to mimic practices at milk banks; however, a variety of time and temperature combinations were tested.Although many studies reported that viruses including Ebola, Marburg, Zika, cytomegalovirus, and HIV appear to be completely inactivated after 30 min at 62.5°C -63°C (Table 1), others report inactivation after a shorter duration; it remains unclear whether Holder pasteurization for shorter times might effectively inactivate these viruses.Arriving at a consensus is difficult given that one study might assess reductions in surviving virus concentrations before and after Holder pasteurization and another might assess at different time points during the pasteurization process.Moreover, high-temperature short-time pasteurization, defined here as pasteurization above 70°C for less than 30 minutes, appears to be as effective as pasteurization at lower temperatures for a longer duration.
Given the limited research in a human milk matrix, the inclusion of studies that assessed viral load or detectable live virus in a range of matrices allowed us to assess a broader scope of viruses belonging to numerous taxonomic families.The matrix may influence the effectiveness of pasteurization by altering how heat is distributed; however, our results suggest that irrespective of matrix, enveloped, compared to non-enveloped viruses, generally require less input of thermal energy in order to achieve similar reductions in viral load or live virus concentration.This suggests that the results presented in Table 2 may, to a certain degree, be representative of how viruses could be inactivated by heat in human milk.In all matrices, including human milk, pasteurization at temperatures of 62.5ºC was generally sufficient to reduce surviving viral load by several logs or to below the limit of detection-depending on the starting concentration of virus and whether it was enveloped.To completely inactivate non-enveloped viruses, such as bovine viral diarrhea virus, hepatitis A or porcine parvovirus in human milk or in other matrices, temperatures above 63ºC (70ºC -90ºC) or a significantly longer duration at 60ºC-63ºC (Table 2) is generally required.Overall, the results are consistent with the logarithmic thermal death time curve where the same degree of thermal lethality can occur at varying temperatures depending on holding time; pasteurization at higher temperatures for shorter durations or lower temperatures for longer durations yielded similar results in terms of the magnitude of infectivity reductions.
Finally, while we cannot discount any differences in response to thermal pasteurization, viruses that were tested in both a human milk and non-human milk matrix appeared to require similar temperatures to elicit a given log reduction in infectivity.Nevertheless, there was significant variability in the duration of pasteurization tested, making it difficult to draw any conclusions as some viruses may require greater time at temperature for one matrix, and less time at temperature for another.In addition to there being a wide range of matrices included as part of the nonhuman milk group, differences in the time may be an artefact of the design of the respective studies; in many cases, viral infectivity or load was not always assessed longitudinally, but after a predetermined length of time.Consequently, this may overestimate the amount of time required to achieve a certain degree of inactivation, making it difficult to compare and aggregate the results from different studies.
There are many strengths of this rapid review.First, we carried out a robust search strategy, in addition to manual searches of grey literature, to generate a complete list of studies, irrespective of language or year published that assessed the impact of thermal pasteurization on viral load in human milk and other matrices.The studies in this review reported on a wide range of thermal pasteurization parameters (low-temperature long-time, high-temperature short-time) across several viruses in a diverse set of matrices.Despite these, the interpretation of our results should be considered alongside its limitations.First, this review was conducted by a single reviewer which may have introduced potential selection bias during initial screening.As a result, our review may not have captured all possible studies.Despite this, the purpose of this review was to rapidly and broadly characterize how viruses in any matrix, including human milk, might respond to thermal pasteurization.Second, the reduction in viral load or detectable live virus that was extracted was approximated if multiple strains of a given virus genus were studied, despite the potential of strain-specific variation in thermal resistance.Third, in our comparison of studies that assessed similar viruses in both a human milk and non-human milk matrix, we chose to aggregate the results to match, to the best of our ability, the pasteurization parameters tested in human milk.While this may have allowed us to assess the temperature and time requirements to achieve a certain log reduction, we were limited to a narrow range of pasteurization conditions.
To our knowledge, this rapid review is the first to broadly summarize the literature that has reported on the impact of any thermal pasteurization on virus survival.The results from this study highlight our limited understanding with respect to the effect of thermal pasteurization on viruses in human milk-this is especially relevant given the possibility that novel viruses, including SARS-CoV-2, may be present in human milk.Although currently, there is insufficient evidence to suggest that SARS-CoV-2 is expressed in milk and could lead to vertical transmission, it may also be present as a contaminant (Lackey et al. 2020).Based on the literature review, Holder pasteurization (62.5°C, 30 minutes) may be sufficient to inactivate nonheat resistant viruses that may be present in HM, including coronaviruses.Though our attempt to rapidly survey all known viral families may help provide some insight into how novel viruses may respond to thermal pasteurization, additional research is warranted to synthesize empirical evidence using human milk as the matrix.

Figure Caption Figure 1 .
Figure Caption Figure 1.PRISMA flow diagram describing the selection of studies for inclusion in the review

Table 1 .
Summary of studies assessing the effect of heat, including Holder pasteurization, on viral inactivation in human milk Human immunodeficiency virus, HIV; Immunofluorescence, IF; Green fluorescence protein, GFP; plaque forming unit, PFU; Peripheral blood mononuclear cell, PBMC; Plaque reduction assay, PRA; Reverse transcriptase, RT; Secreted embryonic alkaline phosphatase, SEAP; Tissue culture infectious dose 50, TCID 50 .Complete inactivation refers to a viral load that is below the detectable limit of the assay, unless otherwise noted. .Note..

Table 3 .
Comparing the log reductions in detectable live viruses pasteurized in both a human milk and a non-human milk matrix