2. Materials and Methods
An investigation, combining field measurements and modelling, was used to assess the potential risk for allergic populations exposed to milk applied as a PPP. In regard to hazard identification, the hazard or exposure of interest is bovine proteins contained in skimmed milk (maximum 0.5% fat content) applied to vineyards by helicopter dispersion.
2.1. Field Measurements
The field measurements were carried out on 7 June 2018, between 11:00 a.m. and 1:00 p.m., during the skimmed milk treatment of a vineyard located in the Lavaux region (canton of Vaud, Switzerland, 46°29′57.8″ N 6°42′23.5″ E). The treated area is shown in
Figure 1a. This region is very steep, which makes access to vehicles in the vineyards difficult and makes treatment by helicopter attractive despite its cost. Treatment of the area took about 2 h, during which time the helicopter treated about 50 hectares, flying back and forth between the treated vineyards and the refill area (marked H on the map). The helicopter used on the day of the treatment was a AS350 B2 (Airbus Helicopters, France).
Field measurements were carried out in a vineyard of about 4000 m
2, located in the middle of the treated area, at a volunteer winegrower’s property (
Figure 1b). Quantitative field-based sampling of air was collected at 15 locations (12 inside the treated perimeter, 3 outside). Airborne concentration was measured inside the vineyard perimeter to provide proxy estimates for worker exposure. Measurements outside the vineyard were taken to represent the exposure of the general population under unfavorable conditions (living area adjacent to the vineyard, public road). Outside measurements were taken. Residual milk was also measured on the surface of leaves over 7 days following spraying in order to quantify temporal degradation. A more detailed view of the sampling area, numbering the sampling locations, is provided in the
Supplementary Material (Figure S1).
2.2. Airborne Concentrations
Aerosol samples were taken during the treatment of the selected vineyard. The pumps were laid out in the field as described in
Figure 1b. The three measurement points outside the treatment zone were intended to assess the dissemination of the products by the wind, the helicopter not having the right to spray close to the houses ensuring indirect contact of these samples with the treatment product. The sampling started immediately prior to the treatment and for a period of about 160 min.
Air samples were taken using SKC pumps (224-PCXR4, SKC Inc., Eighty Four, PA, USA) set at a flow rate of 2L/min and equipped with 25 mm IOM Multidust samplers (SKC Inc., PA, USA) adapted to measure the inhalable fraction of aerosols in the air. A fiberglass filter 25 mm 1.0 micron (N° 225–702, SKC Inc., PA, USA) in diameter was used as a sampling medium to collect non-volatile milk material.
In the laboratory, the fiberglass filters were extracted with 2 mL of milliQ water. The vials were put in a bath at 60 °C for 30 min and stirred for 10 min with a rotary shaker. The solutions were filtrated and injected in a ICS-5000 ion chromatography system (Thermo-Dionex, Grand Island, NY, USA) equipped with a DP 5000 pump, an autosampler AS-AP and a thermostatized compartment DC 500 with an electrochemical detector. The analytical column was a carbopac PA1 (50 + 250 mm × 2.0 mm, Dionex Grand Island, NY, USA), maintained at a temperature of 30 °C. The mobile phase was 150 mM NaOH at a flow rate of 0.4 mL/min in isocratic mode. The system was calibrated using a stock solution of 1.5 g Lactose/L, obtained in dissolving lactose standard from Sigma Aldrich (Buchs, Switzerland) in water milliQ. Lactose was used as a marker of milk concentration in the air and on the leaves. This marker was chosen because of the sensitivity and reliability of its determination in ion chromatography. The lactose concentrations obtained were then converted to milk protein concentration according to Equation (1).
Direct-reading air concentrations were also conducted in two locations during the treatment period. Two nephelometers (pDR-1000, Thermo Electron Corp., Beverly, MA, USA) were placed in the treated area and outside the treated area, in parallel with filter samples (
Figure 1b). These devices record the aerosol concentration in the air, continuously, as a function of time. Optical measurements from nephelometers are not specific to a particular substance and must, therefore, be adjusted to the cow milk protein concentration using the results of the filter samples obtained at the same location. They are, therefore, essentially semi-quantitative measures, which are not subject to statistical analysis. They allow visualizing a profile of concentration during the treatment, in order to appreciate the persistence of the aerosols during the treatment and during the hours following the passage of the helicopter.
2.3. Surface Contamination
Grape leaves were harvested immediately after treatment and at successive intervals on the days following treatment (D, D + 1, D + 3 and D + 7). A sample at D + 10 was planned, but a new treatment was given that day. The same sampling locations as for air measurements were used for leaves sampling (
Figure 1b). The leaves were chosen randomly, without paying attention to the presence of droplets or white spots. The criterion was to choose leaves from the upper part of the plant and medium in size. Several people participated in the harvest of the leaves introducing a possible effect of involuntary selection.
The collected leaves were put in plastic bags, identified by date and location and kept in the fridge. Residues on the grape leaves were extracted directly in the collection bags with 40 milliliters of milliQ water. The bags were put 30 min in a 60 °C bath prior to filtration and injection, using the same analytical method as that used for filters. After scanning the leaves for surface determination, the amount of lactose measured on the leaves was converted to an amount of cow milk proteins per unit area to determine the persistence of milk residue as a function of time on the surface of the leaves and its geographical spread during treatment.
Surface contamination far from the treated area (downwind concentrations) was estimated through computer-based modeling using AgDRIFT
® software (version 2.1.1, US EPA, Washington, NW, USA). AgDRIFT
® is a drift model used to assess the movement and deposition of pesticides sprays. It is primarily used as a decision-making tool for low-flight applications. The model requires a list of physical input parameters that are used to calculate evaporation of the liquid phase from particles and to account for wind speed, meteorological conditions, the spraying rate and the helicopter/airplane type and characteristics. The parameters used in this study are given in
Table 1. Realistic conservative estimates were used for most parameters, but ranges were considered for two parameters (boom height and wind speed) which may have varied significantly during treatment.
4. Discussion
This study assessed whether milk applied as a PPP is a risk for allergic populations, namely, children who are most likely to suffer from the most adverse effects. This question is particularly relevant in regions where vineyards are located in tourist areas or close to inhabited areas, where it is difficult to prevent the presence of the public near the treatment area.
Aerosol sampling revealed an estimated average concentration of 0.47 and 0.16 µg/m3 in milk protein inside and outside (but close) the vineyard, respectively. It should be noted that the difference between the two zones is essentially due to the high variability observed in the inner zone of the vineyard. These results are consistent with direct-reading measurements, which showed an increasing aerosol concentration near the treated area, with a maximum of about 0.4 µg/m3 about 45 min after the start of treatment. The air concentration at a distance from the vineyard is difficult to estimate due to the complex geography of the area (steep slope). Simulations conducted with AGDrift, which essentially calculates surface deposition, however, suggested an exponential decrease, with a strong abatement during the first 100–200 m. It is, therefore, reasonable to estimate that the order of magnitude of milk protein concentration, during and within one hour after treatment on public areas in the close vicinity of the treated area (roads, paths and outside dwellings), is within an (conservative) order of magnitude of 0.1–0.5 µg/m3 (or ng/L-air).
The threshold dose of inhaled milk protein was not established. However, it is well known that the inhalation of cow’s milk can induce severe anaphylactic reactions in allergic individuals [
15]. Uncertainty in the dose response makes it difficult to fully characterize the potential health impacts for exposure levels presented in the study. Nevertheless, evidence from the literature has shown that repeated exposure to low doses of inhaled milk proteins might also exacerbate chronic airway inflammation and lead to poor asthma control in the absence of acute immediate reactions [
16]. Considering the high vascularity of the respiratory tract mucous membranes and lack of protective mechanisms unique to the gastrointestinal tract, such as digestive enzymes and low pH, one could hypothesize that smaller doses of inhaled food allergen would induce an allergic reaction compared with an ingested allergen in highly sensitized/allergic individuals [
16]. It is also important to consider that while inhalation is commonly recognized as symptom provoking in individuals who have previously developed food allergy, inhalation can also cause de novo sensitization [
8], especially in occupational settings. Moreover, it is difficult to disentangle the distinct response from the entry route given that particles inhaled through the mouth will impact the back of the throat, and may be subsequently ingested [
11].
Another factor that may contribute to a lower threshold for inhaled food allergens is that the allergenicity of milk proteins may be enhanced by the formation of lactose-protein complexes. Glycation of milk proteins occurs during heat treatment (Maillard reaction), leading to significant changes in the three-dimensional structure of these proteins. These conformational modifications might lead to large glycoprotein complex formation and potentially enhanced allergenicity, depending on certain factors. Intradermal skin test reactivity to β-lactoglobulin–lactose conjugates has been shown to increase by 10- to 100-fold compared with native β-lactoglobulin [
17]. Nowak-Wegrzyn (2004) previously hypothesized that the random distribution of such large complexes may be responsible for differences in allergic reactions [
16]. However, the effects of glycation and subsequent formation of aggregates do not always result in increased allergenicity, and this factor is largely dependent on the type or family of proteins. Caseins naturally tend to form ordered aggregates, which contributes to maintaining their IgE-binding capacity. However, when caseins form aggregates with other proteins, such as whey and wheat proteins, their IgE-binding capacity is increased or reduced, respectively [
1].
In regard to the surface concentrations of lactose on the leaves of the plant, our results indicate that the amount of milk proteins after treatment inside and close to the treated area was of 10–100 μg/100 cm2. Milk proteins were still present on the surface of the leaves several days following spraying. Three days after treatment, they became very low, either due to leaf absorption or natural degradation. The simulation results indicate that surface contamination downwind of the vineyard could remain within the same order of magnitude of 100–200 m.
Taken together, these results demonstrate that exposure to airborne milk proteins concentrations, as well as those on the surface of the leaves, in and around treated agricultural areas is not negligible. This indicates a risk both for inhaled as well as ingested exposures. Previous reports have described children with severe milk allergy having acute allergic reactions after the ingestion of food products containing >10 parts per million of total milk protein. Moreover, foodborne allergies can be triggered by infinitesimal doses (from micro- to nanogram), irrespective of the route of exposure [
18]. It is important to note that those who are allergic to food by ingestion may react to the same food by inhalation.
Risk Characterization
The process of risk characterization considers the route of exposure, and assesses the probability of exposure, matched with the severity or the impact of that exposure. In this study, we can hypothesize that the likelihood of exposure will be low, but that the severity or health impact is potentially high.
A shown in
Table 2, the risk by inhalation and ingestion can be considered as non-negligible. However, it is important to note that the time window of exposure opportunity can be very different for inhalation exposure and surface contact exposure for ingestion. The fact that exposure may occur outside the treated area or several hours or days after the actual treatment is a concern in tourist areas or near populated areas. Usual safety measures, which include prohibiting entry into the treatment area, similarly to any chemical treatments by helicopter or airplane, do not eliminate this risk. Considering that it only concerns a minority of the population, all-public measures, which would have to be taken several hundred meters from the treated area and several days after treatment, do not seem realistic. Targeted information messages to vulnerable populations, for example, through allergists, could be more effective and make skimmed milk treatments safer, as they are an interesting alternative to traditional, more polluting pesticides that are dangerous for humans and any other living thing (bees, birds, etc.).