Epidemiologic studies have demonstrated a clear association between exposure to airborne particles at concentrations commonly found in major metropolitan areas worldwide and increased mortality rates [1, 2]. Ambient airborne particulate matter (PM) is ubiquitous; however, its concentrations, particle sizes, and chemical compositions can vary substantially across locations and over time [3]. The aerodynamic diameter is defined as the diameter of a spherical particle with a density of 1 g/cm³ [4]. PM with an aerodynamic diameter of less than 10 µm (PM10) is further classified into size-based categories: coarse particles (2.5–10 µm), fine particles (PM2.5, <2.5 µm), and ultrafine particles (UFP or PM0.1, <0.1 µm or 100 nm).

Most studies have relied on data obtained from environmental monitoring, which provides valuable information for assessing regulatory compliance in pollution control but is a poor surrogate for measuring the full extent of particulate matter’s adverse effects on the human respiratory system. To better understand these impacts, it is important to consider measurements of a pollutant’s internal dose, which account for the route of exposure (e.g., inhalation, dermal absorption) and variations in the delivered dose at the individual level [5]. The internal dose of pollutants can be accurately assessed only through biological monitoring. In this context it was shown that exposure to metal (loid)s was evaluated by measuring the levels of aluminum, arsenic, cadmium, chromium, mercury, and nickel in the urine of 100 participants from urban residential areas in Iran [6]. Tear fluid has also been reported as a useful matrix for biomonitoring environmental and chemical exposures [7]. Biomonitoring of particulate matter (PM) and polycyclic aromatic hydrocarbons (PAHs) was conducted among children attending schools in polluted urban and industrial areas. These children, exposed to higher levels of PM and PAHs, exhibited increased concentrations of urinary PAH metabolites compared to children from rural areas [8].

Assessing particulate burden in the lungs can be performed using a combination of fiberoptic bronchoscopy and bronchoalveolar lavage (BAL). However, the relative invasiveness of this technique has limited its use as a practical tool for screening programs, evaluating exposure levels, and conducting repeated follow-up tests in large populations. The ongoing search for noninvasive techniques has led to the development of methods such as analyzing exhaled breath condensate (EBC) and examining cells, mediators, and particulate matter in samples of induced sputum (IS).

In recent years, our research on induced sputum (IS) and exhaled breath condensate (EBC) has focused on occupational and environmental exposure, specifically using micro-sized airborne particles (Dipa 2000 Analyzer Donners Technology). We have studied several populations, including: (a) workers exposed to hazardous dust in Israel [9, 10]; (b) firefighters exposed to World Trade Center dust in the USA [11]; and (c) asthmatic children in the Tel-Aviv area [12]. We have also conducted ongoing complementary measurements of nanoparticulate matter in IS samples using NanoSight technology (NanoSight LTD, London, UK) [13, 14].

In the present study, we use exhaled breath condensate (EBC), fractional exhaled nitric oxide (FeNO), and metabolome measurement in saliva to biomonitor the Haifa population for exposure to ultrafine particles as a surrogate to induced sputum. During the COVID-19 pandemic, sputum induction was classified as an aerosol-generating procedure (AGP), increasing the risk of pathogen exposure and infection [15].

EBC has long been used to monitor toxic effects [16], and FeNO is a simple, safe, and noninvasive method for detecting airway inflammation [17]. Saliva, also known as oral fluid, is a natural filtrate of blood containing small molecules, metals, proteins, and DNA, making it valuable for omics studies such as Epigenome-Wide Association Studies (EWAS). Due to its “stress-free” collection process, saliva was shown to be useful to asses’ biomarkers [18]. Several reviews were published recently [19, 20] summarizing the role of saliva in biomonitoring.

This study is the first to bio monitor the toxic effects of ultrafine particles in Haifa Bay, a heavily industrialized area where numerous factories operate within a confined space. The resulting accumulation of airborne toxins has created unusually high pollution levels, posing potential risks to the health of the local population.

No differences were observed in demographic characteristics, smoking habits, or functional parameters between the medical staff in Tel Aviv and Haifa (Tables 1,2). However, questionnaire data revealed significantly higher levels of clinical parameters, including atopy (p < 0.001) and the presence of sputum and cough during illness (p < 0.05)—among the Haifa medical staff compared with their counterparts in Tel Aviv (Table 3).

There were no significant differences in the PFT results of the Haifa and Tel Aviv medical staffs. (Data not shown here). The analysis of EBC samples is shown in Table 4. Significant differences were found in EBC UFP size between the medical staff. The mean size was larger in Haifa (231.75 ± 60.19 nm) than in Tel Aviv (166.99 ± 60.22 nm) in the 1st session (p < 0.001) but not in the second session. In addition, the concentration was significantly lower in Haifa in the 2nd session (2.93 ± 1.98 10⁸ particles/mL vs. 5.06 ± 6.02 10⁸ particles/mL in Tel Aviv, p < 0.05).

There were also significant differences in the mineral content (XRF analysis) between the two groups only during the first session. The levels of the metal’s molybdenum, niobium and zirconium were significantly higher for the Haifa medical staff (p = 0.001 vs. Tel Aviv medical staff), while the cadmium and palladium levels were significantly lower for the Tel Aviv medical staff (p < 0.001 and <0.05, respectively, vs. Haifa medical staff). There were no comparable differences during the second session. No differences were found in the EBC pH levels between the two groups in both sessions.

The analysis of saliva samples is shown in Table 5. Differential cell count (DCC) analyses were performed in the saliva and there were no significant group differences in any of the DCC parameters.

The biochemical analysis of the saliva samples showed higher levels in all parameters tested in both Tel Aviv and Haifa during the second session, but the increase in both sessions was significant only in blood urea nitrogen (BUN) and lactate dehydrogenase (LDH). Table 6 displays the negative significant association between the accumulation of particles <2.5 µm, <5 µm and <10 µm in the saliva supernatant and a functional parameter (FVC), indicating that the greater the accumulation of particles, the lower the functional parameters. Testing the association of particle accumulation with functional parameters (FVC and FEV₁) in Tel Aviv and Haifa groups separately indicates that some of the associations lose their significance in the TLV group but not in the Haifa group.

The acidification of airways was measured by pH in EBC. There was a positive significant association of PH with functional parameters (FVC and FEV₁) in the Haifa group, indicating that a low pH is associated with low functional parameters.

Inflammation in airways was measured by FeNO, and positive significant associations between FeNO and UFP size distribution and between FeNO and mean size were revealed, indicating that UFP of all sizes induce elevated values of FeNO.

Table 7 presents the fit indicators of the LCA procedure. The first criterion for a minimum of 20 participants in class ruled out the 3-Class solution.

2 Class solution has a lower AIC and BIC than the 1 Class solution and a significant Lo-Mendell-Rubin likelihood ratio test, therefore we identified the 2 Class model as optimal. The first class consisted of 34 subjects and the second class consisted of 65 subjects (Table 8).

We analyzed the association between the clusters and the demographic and clinical parameters (Table 8), and found that cluster 2 was characterized as having:

Greater proportion of the Haifa population.

This cluster included more Haifa individuals with high concentrations of UFP in EBC. The biochemical analysis of the saliva showed high concentrations of LDH (as a marker of degradation). The inflammatory effect of the higher concentration of UFP was depicted by symptoms of atopy and high percentages of lymphocytes in saliva.

Biomonitoring urban pollution involves using biological systems—such as plants, animals, or microorganisms—to assess the level and impact of pollution in urban areas. We present here the use of different biological samples to monitor pollution in two big cities in Israel.

The population studies in Tel Aviv and Haifa displayed the same PFT values and showed no differences between them when compared individually in the first and second session. These results confirmed those of our earlier studies on other worker populations compared to normal populations [28]. PFT values are not sufficiently sensitive to serve as a tool for biological monitoring.

We therefore collected saliva and EBC as less invasive alternatives to induced sputum.

Saliva, also known as oral fluid, is a natural filtrate of blood that contains various omics features, including small molecules, metals, proteins, and DNA. It offers numerous advantages over other biofluids, as it can be collected safely and noninvasively with minimal training. Additionally, saliva is rich in biological information [18]. The multiple functions provided by Saliva are essential for proper protection and functioning of the body as well as for general health. With the spread of the COVID pandemic, the saliva-based molecular tests have shown a similar sensitivity and specificity compared to nasopharyngeal tests for SARS-CoV-2 [29, 30]. The use of saliva in other pathological conditions, such cancer [31], autoimmune disease [32] and Alzheimer disease [33] have also been reported.

In our current investigation, we demonstrated the validity of our methods using samples of saliva and EBC to measure the mean size of particulate matter in the airways. It is known that inhaled particles distribute in the airways according to their size and become smaller in the deeper sites in the bronchial tree. They deposit efficiently in the human lung and can translocate to the systemic circulation and deposit into various organs [34–36]. It emerged that the mean size of particles in saliva was significantly larger than those measured in EBC (data not shown here) since the saliva sample recovered from an upper airway compartment. These results agree with those shown by us in a previous study where we showed that particles recovered from bronchioalveolar lavage are smaller than those recovered from induced sputum [37]. The comparison of these very significant changes in the size of UFP particles between the two populations at two periods of time may indicate that this parameter can be very sensitive to some unpredictable atmospheric changes like those resulting from lockdown that can act as a confounder. This was demonstrated in a very recent observational study that showed that reduced levels of air pollution during the COVID-19 lockdown in Israel were reflected in increased levels of UFP airway contents [38].

Interestingly, the large size of particles in Haifa and the high concentration of UFP particles occurred in parallel with high levels of metals during the first session. During the second sessions the size decreased, and the level of concentration lowered by almost one-half (5.06 ± 6.02 vs. 2.93 ± 1.98). This phenomenon can be explained by the fact that metals attached to the surfaces of particles exhibit enhanced properties due to their small dimensions and high surface-to-volume ratio [39].

The results of all the biochemical parameters were consistently higher in Haifa compared to Tel Aviv in both the first and second session, but significant for two of them (LDH and BUN). We focused particularly on LDH levels, as this metabolite has been identified as an indicator of cellular integrity disturbances and a marker of lung and pulmonary endothelial cell injury [40]. Blood urea nitrogen (BUN) is a serum byproduct of protein metabolism. It is one of the oldest prognostic biomarkers of heart failure. Urea is formed by the liver and carried by the blood to the kidneys for excretion and may be used as a marker for predicting renal disease.

Accumulation of PM in saliva samples (PM 2.5 µm, PM 5 µm, and PM 10 µm) were negatively associated with PFT parameters (FVC and FEV1), indicating that the higher the accumulation of this fraction of PM the worse will be the PFT parameters. This correlation remained significant only in the Haifa population when the two urban populations were compared. The fact that inhaled particles correlated with worse PFT results had already been reported in elderly [41] and in a pediatric population [42]. While those studies used measurements in the environment PM, however, to the best of our knowledge ours is the first report on PM findings in Saliva.

A negative association was found between pH and PFT, demonstrating that acidification of the airways (i.e., lower pH) was correlated to worse PFT results supporting our earlier findings [28].

FeNO was used in this study as a surrogate for eosinophils in IS and its concentration was shown to be positively associated to the accumulation of UFP. This was significant only in the Haifa study population. These results may indicate that small particle accumulation in the airways as measured in EBC caused high inflammatory values of FeNO in the Haifa population.

Epidemiological and toxicological research support a link between air pollution and an increased incidence and/or severity of airway inflammation. FeNO is a simple, safe and noninvasive method to detect airway inflammation. It is also correlated well with eosinophil count and eosinophil cationic protein in induced sputum and serves as a widely used metric for the evaluation and management of airway inflammation [43–45].

In this context toward this end, we performed a latent class analysis (LCA) to identify the number of homogenous subgroups according to selected measurements within the overall sample (detailed in the Statistical Methods section). Once the ideal number of classes had been determined, individuals were assigned to their most likely class and compared between the classes based upon demographic, clinical, and biological variables by means of a univariate analysis.

The results yielded two homogeneous clusters. The measurements chosen as indicators of the LCA were: FEV₁(the only physiological parameter), FeNO, UFP (very small particles mean size) pH, cadmium, palladium, molybdenum, niobium and zirconium in EB, BUN, LDH in saliva. These markers were selected according to the parameters that had been shown to be with significant differences and sensitive all over the study performed.

Cluster 2 included more Haifa individuals who had high concentrations of UFP in EBC. The biochemical analysis of the saliva showed high concentrations of LDH (as a marker of degradation). The inflammatory effect of the higher concentration of UFP was reflected by symptoms of atopy and high percentages of lymphocytes in saliva.