Effects of exposure to PM_2.5 and its chemical components on survival of patients with lymphoma

Abstract

Background

There is accumulating evidence of an association between particulate matter smaller than 2.5 μm (PM_2.5) and cancer. However, the impact of air pollution on the survival of lymphoma patients has received little attention.

Methods

This retrospective study was conducted at Henan Cancer Hospital. We included lymphoma patients admitted between January 2017 and December 2020, with follow-ups conducted via telephone. Daily gridded data on PM_2.5 and its chemical components were obtained over a 10 km × 10 km grid. To assess the association between PM_2.5 levels and lymphoma mortality, we employed a Cox proportional-hazards model. Additionally, a restricted cubic spline function was used to evaluate nonlinear associations between chemical components and 1- and 2-year PM_2.5 exposure levels with lymphoma survival. Finally, a multi-pollutant analysis using quartile-based g-calculation was performed to estimate the joint association of PM_2.5 and chemical component exposure with mortality.

Results

The 1-year PM_2.5 exposure level was positively associated with the risk of lymphoma death, as were its chemical components. In contrast, the 2-year PM_2.5 exposure level showed no significant association with lymphoma mortality, although its components exhibited a positive correlation. In a multi-pollutant joint effects analysis, 1-year exposure to organic matter (OM) was associated with lymphoma mortality (HR: 1.56; 95% CI: 1.01–2.40), while 2-year exposure to black carbon (BC) was also associated with increased mortality risk (HR: 1.82; 95% CI: 1.10–3.02).

Conclusions

Exposure to PM_2.5 may adversely affect the survival of lymphoma patients. Further studies, including mechanistic investigations and clinical trials, are needed to validate this association. Additionally, lymphoma patients may benefit from residing in areas with good air quality after diagnosis, as this could potentially improve their survival outcomes.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12889-025-25877-1.

Background

Air pollution, a major global health concern, has been exacerbated by rapid worldwide industrialization [1]. In 2019, data indicated that approximately 99% of the global population resided in regions failing to comply with the World Health Organization’s air quality guidelines, raising significant international concern [2, 3]. Air pollution consists of a complex blend of particulate matter, carbon monoxide, lead, nitrogen dioxide, ozone (O₃), and sulfur dioxide [4]. Of particular concern is PM_2.5, fine particulate matter smaller than 2.5 μm, which is notorious for its adverse health effects. Globally, ambient (outdoor) air pollution is implicated in an estimated 4.2 million premature deaths [5]. Substantial evidence has established the association between PM_2.5 exposure and increased mortality risks from respiratory [6, 7], and cardiovascular diseases [8–10].

Lymphoma encompasses a diverse array of malignant lymphocyte neoplasms that may affect lymphatic tissue, bone marrow, and extranidal sites [11]. Mature lymphoid neoplasms are broadly classified into three main categories: mature B-cell neoplasms, mature T-cell or natural killer (NK)-cell neoplasms, and Hodgkin lymphoma. Among these, B-cell and NK-cell neoplasms fall under the umbrella of non-Hodgkin lymphomas, which encompass over 60 distinct subtypes [12]. In 2019, it was estimated that more than 82,000 new lymphoma cases were diagnosed in the United States, accounting for 4.7% of all new cancer cases [11]. The current 5-year survival rates are 72.0% for non-Hodgkin lymphoma and 86.6% for Hodgkin lymphoma [11]. Approximately 21,000 deaths from lymphoma were found in 2019, accounting for 3.5% of all cancer fatalities [11]. Hvidtfeldt et al. observed a heightened multiple myeloma risk with increased PM_2.5 exposure [13]. The National Health Interview Survey identified positive correlations between exposure to PM_2.5 and mortality from Hodgkin lymphoma, non-Hodgkin lymphoma, and leukemia [14]. Furthermore, the American Cancer Society Cancer Prevention Study II Nutrition Cohort highlighted detrimental links between outdoor benzene exposure and T-cell lymphoma, myelodysplastic syndromes, and follicular lymphoma in men [15]. Comprehensive case-control studies across Denmark’s entire adult population also indicated an elevated risk of acute myeloid and chronic lymphoblastic leukemia related to PM_2.5 exposure [16, 17]. However, these investigations share critical limitations: no study has specifically examined PM_2.5 chemical constituents’ effects on lymphoma mortality, nor have they adequately accounted for clinical variables, including disease stage and treatment modalities - factors known to influence survival outcomes substantially [16, 18].

Notably, emerging evidence supports the biological rationale for focusing on shorter exposure windows in cancer progression. Mechanistic studies demonstrate that even brief PM_2.5 exposure, as short as 24 h, can activate oncogenic pathways [18], such as through aryl hydrocarbon receptor (AhR) nuclear translocation and Interleukin-18(IL-18) production, which are well-established drivers in lymphoma pathogenesis and progression. AhR activation has been shown to play a carcinogenic role in lymphoma [19, 20], often mediated by effectors like IL-18 [21], and clinical data link elevated AhR pathway activity and serum IL-18 levels to poorer survival in lymphoma patients [22]. This underscores the relevance of investigating recent PM_2.5 exposures in lymphoma outcomes.

The primary objective of our study was to explore the relationships between exposure to various components of PM_2.5 and the survival outcomes of patients with lymphoma at Henan Cancer Hospital. By evaluating the impact of PM_2.5 exposure on components related to the survival of patients with lymphoma, we hope not only to enhance patient care but also to provide a theoretical basis for public health strategies addressing health issues induced by PM_2.5 exposure. This inquiry is vital for understanding the ways in which specific elements of air pollution affect lymphoma prognosis and for providing targeted interventions to mitigate these effects.

Methods

Study population

This prospective cohort study was conducted at the Henan Cancer Hospital in central China. Patients hospitalized between January 1, 2017, and December 31, 2020, were selected based on the following criteria: (1) a lymphoma diagnosis according to the International Classification of Diseases, 10th Revision; and (2) having received inpatient treatments, including surgery, chemotherapy, and radiotherapy. Patients without a current residential address or those without residential addresses in Henan were excluded (Fig. 1). Patients from other provinces were excluded due to higher heterogeneity and less stable residential addresses, which could affect data consistency. Socioeconomic and clinical treatment data were extracted from electronic medical records. Due to the geographic nature of China’s health insurance policies and the regional patient base of our hospital, our internal data show that about 98% of the patients were residents of Henan Province. This demographic feature supports the interpretation that our cohort effectively represents a population with long-term exposure to ambient PM_2.5 in Henan, substantially reducing the risk of exposure misclassification related to short-term residence.

Fig. 1.

Study flowchart

The study design was approved by the institutional ethics committee of the hospital (2021-KY-0054-001), and written informed consent was obtained from all participants before initiation of the follow-up process. The reporting of this study adheres to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines.

Follow-up

Patient follow-ups were conducted via telephone every 3 months for the first 2 years post-treatment, biannually up to the 5th year, and annually thereafter. During these follow-ups, information regarding each patient’s status (alive or deceased), date of death (if applicable), and date of last follow-up was recorded. Follow-up time is defined as the period from the date of disease diagnosis to the date of death (or the date of last follow-up). Patients were followed from initial diagnosis until death or for up to 10 years, whichever occurred first.

Ambient air pollutant exposure

We used daily gridded data on PM_2.5, its chemical components, and ozone (O₃) over a 10 km × 10 km grid sourced from the China Air Pollution Tracking Database (TAP). This near-real-time database integrates data from various studies, including those focusing on PM_2.5, chemical composition analysis, and O₃ variation estimation. The PM_2.5 prediction model, extensively documented in previous research, amalgamates PM_2.5 observations, Community Multiscale Air Quality (CMAQ) model simulations, Moderate Resolution Imaging Spectroradiometer Level 2 Aerosol Optical Depth retrievals, meteorological parameters from Modern-Era Retrospective Analysis Version 2, and a two-stage random forest model incorporating land-use information. The initial model predicts episodes of high PM_2.5 and uses a synthetic minority oversampling technique to address the imbalance between high and normal pollution levels in the training dataset, thereby enhancing predictive accuracy. The second model uses the stage 1 predictions, along with other predictors, to estimate discrepancies between CMAQ and PM_2.5 simulations. Satellite data that were missing owing to cloud cover or high surface albedo were replaced by considering the data availability status as a cut point in the first level of the decision tree. An assessment of the two-stage PM_2.5 prediction model revealed an out-of-bag R² (linear regression R² between measurements and predictions from a tree that did not contain each measurement) ranging from 0.80 to 0.88 [23]. The model shows that the national model has a good prediction effect in different regions, such as Central China (correlation coefficients: 0.9), where this study regions are located. Based on the above-mentioned PM_2.5 model with full spatial coverage, TAP developed a PM_2.5-species concentration dataset. The PM_2.5-species data were extracted using CMAQ simulations and corrected for bias using improved dust modeling and the XG-Boost algorithm. Accurate conversion factors were obtained to derive the concentrations of sulfate, nitrate, ammonium, organic matter (OM), and black carbon (BC) [4]. The chemical composition of PM_2.5 agrees well with the available observations over the study period, with daily correlation coefficients ranging from 0.67 to 0.80 [24]. OM is a complex mixture that includes both directly emitted airborne particles and secondary particles produced by chemical reactions. BC is produced by the incomplete combustion of biomass and fossil fuels.

TAP uses machine learning techniques to predict 10 km × 10 km 8-hour average O₃ concentrations [25]. This prediction process involved the fusion of data from several sources, including O₃ measurements, satellite data, Weather Research and Forecasting and CMAQ simulations, the normalized difference vegetation index, nighttime light pollution levels, and population statistics. Two random forest models were used to compare O₃ measurements to the predictor variables. One model included satellite data, whereas the other model did not. However, there is some uncertainty in the predictions owing to gaps in the available satellite data. An elastic net model was introduced to merge these predictions, resulting in improved accuracy and gap-filling capabilities. In addition, residual data were simulated to refine the predictions using spatiotemporal kriging interpolation. 5-fold cross-validation was used to rigorously evaluate the performance of the algorithm in estimating O₃ concentrations. The patient’s residential address refers to the permanent address of the household as reported during the hospital admission. The patient’s residential address was located within the grid, and the exposure level to air pollutants on that day was matched (Fig. 2).

Fig. 2.

Geographical distribution of the study area and study participants

Patients’ residential addresses were geocoded to geographic coordinates and matched to the corresponding pollution grid cells to assign exposure levels for each year throughout the study period. We defined short-term post-diagnosis exposure as the average PM_2.5 concentration during specific time windows after diagnosis. Specifically, ‘1-year exposure’ refers to the mean exposure level during the first year after diagnosis, while ‘2-year exposure’ represents the mean level during the first and second years after diagnosis. These measures capture recent post-diagnosis exposure rather than prior, lagged exposures.

Covariates

We adjusted for potential confounders according to previous studies. Participant age was treated as a continuous variable, sex (male, female), ethnicity (Han, other), occupation type (mental labor, physical labor, unemployment/retirement), marital status (married, other), operative season (Spring, Summer, Autumn, Winter), pathological type (Hodgkin lymphoma, Non-Hodgkin lymphoma, unknown), pathological staging (I, II, III, IV), radiotherapy (yes, no), chemotherapy (yes, no), operation (yes, no), smoking (current, former, never), drinking (current, former, never), hypertension (yes, no), and diabetes (yes, no) were treated as categorical variables.

Statistical analysis

In the present study, continuous variables were reported as mean (standard deviation [SD]) and categorical variables as frequency (%). The chemical component and 1- and 2-year PM_2.5 exposure levels were described using percentiles and the mean (SD).

The exposure levels of PM_2.5 and its chemical components were separated into three categories according to percentile, and the Cox proportional-hazards model was used to estimate the associations between these categories and lymphoma death. In addition to Cox proportional hazards models, we performed Kaplan-Meier survival analysis to visualize cumulative survival probabilities stratified by tertiles of PM_2.5 exposure. To ensure correct temporal ordering in our analysis of 1- and 2-year pollutant exposure effects on mortality, we excluded participants who died or were lost to follow-up during the exposure period. For mortality risk assessment, we reported hazard ratios (HRs).and 95% CIs and corrected for potential confounders for multifactorial analyses. A restricted cubic spline function combined with the Cox proportional-hazards model, with the 10th, 50th, and 90th percentiles set as knots, was used to estimate the nonlinear associations between the chemical components and 1- and 2-year PM_2.5 exposure levels and lymphoma survival. To estimate the joint association between PM_2.5 and chemical component exposure and mortality, we conducted a multi-pollutant analysis using quartile-based g-calculation, a statistical method for evaluating the combined effects of multiple exposures [26]. Similarly, we assessed the effect of 3- and 6-month exposures on the risk of death in lymphoma patients. We grouped pollutant exposures into quartiles and then executed Cox regression models for sensitivity analyses. We additionally constructed logistic regression models using death as the outcome. Compared to traditional multi-pollutant models with weighted quantile sum regressions, this approach further integrates its estimation process with g-calculations in a marginal structural model, overcoming one-way assumptions and allowing for non-linearities in joint exposures. PM_2.5 and chemical component exposures were included in multi-pollutant analyses, adjusting for the same confounding factors used in the main analyses to estimate their relative contribution weights. 1000 Monte Carlo simulations were used to obtain precise 95% CIs for the point estimates.

Statistical analyses were performed with R software (4.0.3). Two-sided p-values < 0.05 were considered statistically significant.

Results

In this study, of the 2005 patients with lymphoma, 56.76% were male, 67.23% were aged 60 or younger, 87.83% were married, and 91.02% had Hodgkin lymphoma (Table 1). During hospitalization, 20.40%, 92.82%, and 12.26% of the patients underwent radiotherapy, chemotherapy, and surgery, respectively. Additionally, 11.62% and 11.97% were current smokers and drinkers, respectively. Characteristics of participants who died within 1 year of follow-up are described in Table S1, alongside the demographic characteristics of those who were not excluded. As shown in Table S2, a total of 246 deaths occurred during the follow-up period. The longest follow-up duration was 6.45 years, while average follow-up duration was 2.92 years.

Table 1.

Basic characteristics of participants

Characteristics	Total(N = 2005)
Gender, n(%)
Male	1138(56.76)
Female	867(43.24)
Age, n(%)
~ 60	1348(67.23)
61 ~ 70	459(22.89)
71~	198(9.88)
Occupation types, n(%)
Mental labor	357(17.81)
Physical labor	864(43.09)
Unemployment or retirement	784(39.10)
Ethnicity, n(%)
Han	1986(99.05)
Others	19(0.95)
Marital status, n(%)
Married	1761(87.83)
Others	244(12.17)
Diagnostic season, n(%)
Autumn	431(21.50)
Spring	591(29.48)
Summer	537(26.78)
Winter	446(22.24)
Pathological type, n(%)
Hodgkin’s lymphoma	172(8.58)
Non-Hodgkin’s lymphoma	1824(91.02)
Unknown	8(0.40)
Pathological staging, n(%)
I	275(17.11)
II	381(23.71)
III	367(22.84)
IV	584(36.34)
Radiotherapy or not, n(%)
Yes	409(20.40)
No	1596(79.6)
Chemotherapy or not, n(%)
Yes	1861(92.82)
No	144(7.18)
Operation or not, n(%)
Yes	286(14.26)
No	1719(85.74)
Smoking, n(%)
No	1573(78.45)
Quit	199(9.93)
Still	233(11.62)
Drinking, n(%)
No	1683(83.94)
Quit	82(4.09)
Still	240(11.97)
Hypertension or not, n(%)
No	1844(91.97)
Yes	161(8.03)
Diabetes or not, n(%)
No	1914(95.46)
Yes	91(4.54)

As shown in Table 2, the mean PM_2.5 exposure levels at 1 and 2 years after diagnosis were 56.08 µg/m³ (SD: 9.25) and 53.99 µg/m³ (SD: 8.53), respectively. The chemical components of PM_2.5 consist of nitrate, OM, sulfate, ammonium, and BC. The evaluated component exposure levels 1 year after diagnosis, from high to low, were 14.31 (SD: 2.27), 12.66 (SD: 1.95), 10.22 (SD: 1.41), 9.00 (SD: 1.37), and 2.34 (SD: 0.40) respectively; those 2 years after diagnosis, from high to low, were 13.75 (SD: 2.17), 12.19 (SD: 1.78), 9.89 (SD: 1.29), 8.65 (SD: 1.29), and 2.25 (SD: 37), respectively. The PM_2.5 and component exposure levels of deceased participants during the follow-up period were significantly higher than those of the surviving participants (Table S2).

Table 2.

Distribution of PM_2.5 and component exposures after participant diagnosis

Exposure level	Mean (SD)	Percentage distribution
		2.5th	10th	25th	Median	75th	90th	97.5th	Max
1-year exposure level, µg/m3
PM2.5	56.08(9.25)	39.01	44.95	49.58	55.19	63.05	67.56	73.86	89.52
sulfate	10.22(1.41)	7.46	8.53	9.29	10.10	11.25	11.92	12.89	15.96
nitrate	14.31(2.27)	9.60	11.46	12.78	14.20	16.20	17.08	18.12	22.00
ammonium	9.00(1.37)	6.39	7.30	8.00	8.95	10.07	10.66	11.36	14.13
OM	12.66(1.95)	9.26	10.32	11.09	12.55	14.24	15.16	16.31	19.08
BC	2.34(0.40)	1.61	1.88	2.05	2.30	2.65	2.86	3.07	3.63
2-year exposure level, µg/m3
PM2.5	53.99(8.53)	37.86	43.79	48.3	52.88	60.27	65.13	71.33	81.81
SO2- 4	9.89(1.29)	7.36	8.35	9.08	9.73	10.84	11.53	12.40	14.39
nitrate	13.75(2.17)	9.41	11.03	12.3	13.55	15.47	16.61	17.48	19.93
ammonium	8.65(1.29)	6.21	7.04	7.73	8.49	9.68	10.35	10.99	12.74
OM	12.19(1.78)	9.05	10.17	10.77	12.03	13.52	14.55	15.80	17.59
BC	2.25(0.37)	1.56	1.81	1.98	2.20	2.54	2.75	2.97	3.34

Associations between PM_2.5 and component exposure levels and survival were assessed in the middle and highest tertiles, with the lowest tertile serving as the control group. Classification into the three categories was based on percentile. Kaplan–Meier curves showing cumulative survival of lymphoma patients over follow-up stratified by low, medium, and high PM_2.5 exposure (Figure S3).

For the associations between 1-year PM_2.5 and component exposure levels and survival (Table 3), after adjusting for potential confounders, the highest exposure group had been exposed to PM_2.5 (HR: 1.69, 95% CI: 1.12–2.56), sulfate (HR: 1.86, 95% CI: 1.24–2.81), nitrate (HR: 1.90, 95% CI: 1.27–2.86), ammonium (HR: 1.96, 95% CI: 1.28–3.00), OM (HR: 2.39, 95% CI: 1.48–3.84), and BC (HR: 1.85, 95% CI: 1.21–2.83) and had the highest risk of death from lymphoma; The intermediate OM exposure group also had a high risk of death (HR: 2.09, 95% CI: 1.29–3.40). The associations between 2-year PM_2.5 and component exposure levels and survival, after adjusting for potential confounders, were stronger in the intermediate exposure group, which was exposed to PM_2.5 (HR: 1.57, 95% CI: 1.03–2.40), sulfate (HR: 1.69, 95% CI: 1.10–2.59), nitrate (HR: 1.59, 95% CI: 1.03–2.44), OM (HR: 1.84, 95% CI: 1.15–2.93), and BC (HR: 1.60, 95% CI: 1.03–3.10); the highest exposure group had a higher risk of lymphoma death and was exposed to PM_2.5 (HR: 1.75, 95% CI: 1.16–2.63), sulfate (HR: 1.78, 95% CI: 1.18–2.70), nitrate (HR: 1.74, 95% CI: 1.14–2.64), ammonium (HR: 1.74, 95% CI: 1.13–2.66), OM (HR: 1.94, 95% CI: 1.22–3.10), and BC (HR: 1.95, 95% CI: 1.28–2.98). The associations of 1-year exposure to PM_2.5 and its components (analyzed by both quartile and tertile subgroups) with lymphoma mortality risk showed similar patterns (Table S3), whereas the 2-year exposure associations were non-significant (Table S4). Isolated non-linear associations between 1- and 2-year PM_2.5 and component exposure levels and lymphoma survival were assessed using a cubic restricted spline function after adjusting for potential confounding factors. The 1-year PM_2.5 exposure level was positively associated with the risk of lymphoma death (P overall: 0.034, Fig. 3A), as was the level of exposure to the PM_2.5 components (P overall < 0.05, Fig. 3B–F). The 2-year PM_2.5 exposure level was not significantly associated with lymphoma mortality (P overall: 0.055, Fig. 4A), but exposure to its components showed a positive association (P overall < 0.05, Fig. 4B-F).

Table 3.

HR with 95% (CI) for the association of PM_2.5 and component exposure with lymphoma survival

Exposure	HR (95%CI)	HR (95%CI)
1-year exposure (N = 2005)
PM2.5
Tertiles 1	Ref.	Ref.
Tertiles 2	1.18(0.83,1.68)	1.43(0.93,2.19)
Tertiles 3	1.30(0.93,1.83)	1.69(1.12,2.56)
sulfate
Tertiles 1	Ref.	Ref.
Tertiles 2	0.99(0.69,1.41)	1.42(0.92,2.18)
Tertiles 3	1.32(0.95,1.84)	1.86(1.24,2.81)
nitrate
Tertiles 1	Ref.	Ref.
Tertiles 2	1.08(0.76,1.55)	1.39(0.90,2.16)
Tertiles 3	1.42(1.01,1.99)	1.90(1.27,2.86)
ammonium
Tertiles 1	Ref.	Ref.
Tertiles 2	1.31(0.91,1.90)	1.55(0.99,2.42)
Tertiles 3	1.56(1.10,2.23)	1.96(1.28,3.00)
OM
Tertiles 1	Ref.	Ref.
Tertiles 2	1.30(0.89,1.89)	2.09(1.29,3.40)
Tertiles 3	1.61(1.12,2.32)	2.39(1.48,3.84)
BC
Tertiles 1	Ref.	Ref.
Tertiles 2	1.28(0.89,1.84)	1.51(0.98,2.35)
Tertiles 3	1.43(1.01,2.02)	1.85(1.21,2.83)
2-year exposure(N = 1410)
PM2.5
Tertiles 1	Ref.	Ref.
Tertiles 2	1.14(0.80,1.61)	1.57(1.03,2.40)
Tertiles 3	1.24(0.89,1.73)	1.75(1.16,2.63)
sulfate
Tertiles 1	Ref.	Ref.
Tertiles 2	1.13(0.80,1.61)	1.69(1.10,2.59)
Tertiles 3	1.22(0.88,1.70)	1.78(1.18,2.70)
nitrate
Tertiles 1	Ref.	Ref.
Tertiles 2	1.13(0.79,1.60)	1.59(1.03,2.44)
Tertiles 3	1.22(0.87,1.71)	1.74(1.14,2.64)
ammonium
Tertiles 1	Ref.	Ref.
Tertiles 2	1.06(0.74,1.51)	1.40(0.90,2.17)
Tertiles 3	1.25(0.89,1.76)	1.74(1.13,2.66)
OM
Tertiles 1	Ref.	Ref.
Tertiles 2	1.28(0.88,1.84)	1.84(1.15,2.93)
Tertiles 3	1.32(0.92,1.89)	1.94(1.22,3.10)
BC
Tertiles 1	Ref.	Ref.
Tertiles 2	1.26(0.87,1.81)	1.60(1.03,2.47)
Tertiles 3	1.40(0.99,1.99)	1.95(1.28,2.98)

^*Adjusted for age, sex, ethnicity, occupation type, marital status, operative season, pathological type, pathological staging, radiotherapy, chemotherapy, operation, smoking, drinking, hypertension, and diabetes

Fig. 3.

Restricted cubic spline plots of HR for 1-year PM_2.5 and component exposure associated with lymphoma survival. A PM_2.5, B sulfate, C nitrate, D ammonium, E BC, F OM. Models adjusted for age, sex, ethnicity, occupation type, marital status, operative season, pathological type. Pathological staging, radiotherapy, chemotherapy, operation, smoking, drinking, hypertension, and diabetes

Fig. 4.

Restricted cubic spline plots of HR for 2-year PM_2.5 and component exposure associated with lymphoma survival. A PM_2.5, B sulfate, C nitrate, D ammonium, E BC, F OM. Models adjusted for age, sex, ethnicity, occupation type, marital status, operative season, pathological type. pathological staging, radiotherapy, chemotherapy, operation, smoking, drinking, hypertension, and diabetes

As shown in Table S4, Table S6, and Figure S1, three-month exposure to PM_2.5 and its components did not have significant individual and combined effects on mortality in lymphoma patients. As shown in Figure S1 and Table S5, six-month exposure to PM_2.5 had no significant effect on mortality in lymphoma patients, but increased levels of exposure to sulfate, nitrate, ammonium, and BC were positively associated with the risk of mortality in lymphoma patients.

In a multi-pollutant analysis (Table 4), 1-year PM_2.5 exposure was significantly associated with lymphoma mortality (HR: 1.37, 95% CI: 1.15–1.64). Among the components, 1-year OM exposure showed the strongest association (HR: 1.56, 95% CI: 1.01–2.40), followed by 2-year BC exposure (HR: 1.82, 95% CI: 1.10–3.02). Logistic regression model results were consistent with the Cox proportional hazards model in Table S7.

Table 4.

Associations between exposure to PM_2.5 component mixtures for lymphoma survival using quantile-based g-computation

Exposure	HR*
1-year exposure
Mixture effect	1.37(1.15,1.64)
sulfate	1.03(0.58,1.84)
nitrate	1.34(0.76,2.37)
OM	1.56(1.01,2.40)
BC	0.91(0.55,1.50)
ammonium	0.95(0.51,1.78)
2-year exposure
Mixture effect	1.16(0.96,1.41)
sulfate	1.13(0.63,2.03)
nitrate	0.93(0.53,1.64)
OM	0.99(0.62,1.58)
BC	1.82(1.10,3.02)
ammonium	0.85(0.42,1.71)

^*Adjusted for age, sex, ethnicity, occupation type, marital status, operative season, pathological type, pathological staging, radiotherapy, chemotherapy, operation, smoking, drinking, hypertension, and diabetes

Discussion

While the adverse effects of PM_2.5 on respiratory and cardiovascular mortality are well-established [6–10], its heterogeneous impact on lymphoma prognosis remains underexplored, even though lymphoma accounts for 3.5% of global cancer-related deaths [11]. Recent studies reveal divergent spatial patterns: A U.S. cohort analysis reported elevated mortality risks for Hodgkin (HR: 4.18, 95% CI: 1.20–14.60) and non-Hodgkin lymphoma (HR: 1.48, 95% CI: 1.10–1.98), though statistical significance dissipated after multiple testing correction [14]. Conversely, a Chinese DID study demonstrated robust PM_2.5-lymphoma mortality associations (RR:1.106, 95% CI: 1.049–1.168) persisting through sensitivity analyses [27]. These discrepancies may originate from three critical variables: PM_2.5 compositional heterogeneity—with Chinese urban aerosols exhibiting higher absolute BC concentrations (1.885 µg/m³ vs. 1.48 µg/m³ in U.S. cohorts) but lower BC mass fractions (5.2% vs. 11.5% of total PM_2.5 mass) but significantly higher OM contributions (22.5% vs. U.S. levels) [28, 29]—combined with methodological variance in exposure assessment (individual-level geocoding vs. population-weighted regional proxies) and subtype-specific susceptibility patterns potentially masked in aggregated analyses, such as follicular lymphoma’s inflammatory sensitivity to airborne toxicants [16]. Our findings contribute to current understanding by providing component-resolved PM_2.5 characterization while controlling for clinical staging and treatment modalities—key prognostic confounders neglected in prior epidemiological designs. Our study identified significant associations between 1-year exposure to organic matter (OM) (HR:1.56) and 2-year exposure to black carbon (BC) (HR:1.82) with increased lymphoma mortality, suggesting two plausible mechanistic pathways. First, OM—a complex mixture in PM_2.5—likely contains polycyclic aromatic hydrocarbons (PAHs) or other AhR ligands [30]. BC, with its high surface area, may enhance OM bioavailability, facilitating AhR activation in lymphoid tissues [31]. AhR signaling is known to promote regulatory T-cell (Treg) expansion and suppress antitumor immunity [31]. Second, BC itself acts as both a carrier and a direct effector: transition metals (e.g., Fe²⁺, Cu²⁺) adsorbed onto BC surfaces can catalyze Fenton reactions, generating hydroxyl radicals that disrupt mitochondrial electron transport chains and induce oxidative DNA damage [32]. This process may trigger chronic inflammation via mtDNA release and cGAS-STING pathway activation, as evidenced by elevated oxidative stress markers (e.g., 8-OHdG) in lymphoma tissues [33]. The stronger association of BC with 2-year mortality (compared to 1-year for OM) aligns with its potential for cumulative mitochondrial toxicity, which could impair chemosensitivity over time. These pathways may synergize—AhR-mediated immunosuppression creates a permissive microenvironment, while BC-driven oxidative stress directly fuels genomic instability and inflammatory B-cell proliferation, collectively exacerbating lymphoma progression. A key strength of this study was the use of near-real-time PM_2.5 and O₃ data from the TAP database in China. Incorporation of the Baidu Maps API to convert residential addresses into coordinates enabled the precise estimation of daily PM_2.5 and O₃ exposure at an individual level. Moreover, we meticulously considered critical factors such as stage and pathological classification, which, along with behavioral trends such as smoking and alcohol consumption, are crucial for predicting patient survival. The wide geographical scope of the Henan Province, with its areas of diverse air quality ranging from severely polluted to exceptionally clean, significantly enhanced the relevance of our findings.

However, this study did have some limitations. The limited duration of follow-up may not be long enough to adequately observe the outcomes of interest, as longer follow-up could reveal additional associations or long-term effects. Notably, emerging evidence supports the biological relevance of shorter exposure windows in lymphoma, as even brief PM_2.5 exposure can rapidly activate oncogenic pathways such as AhR and IL-18, which are linked to disease progression and survival outcomes [22]. This rationale strengthens the validity of our study design despite the follow-up duration. Data sourced from a single hospital limited the sample size, potentially affecting the robustness of our findings. However, we used individual-level data rather than municipality-level aggregated data, which addressed the issue of diluting the exposure estimates and skewing the risk estimates toward the null. Additionally, we considered individual-level confounders, including pathology, smoking, drinking habits, hypertension, and diabetes, which enhanced the statistical power and accuracy of our study.We acknowledge that our study period partially overlapped with the COVID-19 pandemic, which may have influenced air pollution exposure patterns due to behavioral changes. While we did not explicitly adjust for pandemic-related confounding in our analysis, most of the follow-up period occurred before the pandemic. Furthermore, any behavioral changes during 2020 would likely have led to reduced exposure misclassification, likely reducing bias in the estimates and preventing exaggerated associations. A further limitation was our inability to account for residential mobility, which may affect long-term exposure assessments. Additionally, we lacked data on indoor air pollution and non-residential exposures, potentially affecting exposure estimates. Furthermore, our multi-pollutant analysis included inorganic aerosols and carbon species but did not encompass the detailed composition analysis of certain inorganic elements, such as heavy metals. The absence of data on these specific inorganic compounds represents a limitation, as they contribute to a portion of PM_2.5 and can have impacts on the overall composition and effects of PM_2.5. However, the total PM_2.5 concentration data, which was included in our analysis, ensures that the overall exposure to PM_2.5 is accounted for. Therefore, while the lack of specific inorganic compound data might limit the detailed composition analysis, it does not affect our findings on the impact of PM_2.5 on the survival prognosis of lymphoma patients. The exclusion of individuals who died within the first year of follow-up could potentially lead to an underestimation of exposure effects. However, this approach was primarily taken to maintain the temporal order of causal relationships, i.e., ensuring that the exposure assessment period precedes the outcome, in this case, death. While we acknowledge that this potential bias may have influenced the estimation of exposure effects in our study, we made this design choice to evaluate the long-term exposure effects as rigorously as possible within the constraints of our study design. A further limitation is the potential for misclassification bias, as we only had access to overall survival data and not cause-specific mortality or progression-free survival data. This means that deaths from causes unrelated to lymphoma progression could not be distinguished, which might affect the accuracy of our findings. Although we adjusted for key confounders in multivariate analysis, the lack of cause-specific information prevents definitive conclusions about lymphoma-specific outcomes. We recommend that future studies collect detailed cause-of-death data to more precisely delineate the relationship between air pollution exposure and lymphoma-specific survival. Another limitation is that our data did not permit a reliable classification of participants’ urban–rural residence, so residual urbanicity confounding may remain.

Our findings may have several important implications for public health. Given the observed association between PM_2.5 exposure and lymphoma mortality, future research should include larger sample sizes, longer follow-up periods, and progression-free survival (PFS) times to better determine the impact of PM_2.5 exposure on disease progression. Using PFS times can help eliminate the influence of other competing risks. Additionally, because our component analysis did not include inorganic substances such as heavy metals, future studies should collect more detailed component data to analyze which specific PM_2.5 components affect patient prognosis. Our study suggests that BC and OM may play a role in the progression of lymphoma. This indicates that future research could explore the mechanisms by which these components influence lymphoma progression through cellular experiments and animal models. Our study found that reducing PM_2.5 exposure might extend the survival time of lymphoma patients. Therefore, protecting the environment and lowering PM_2.5 concentrations in the air could not only reduce the incidence of cardiovascular and respiratory diseases but also have a positive impact on the survival prognosis of lymphoma patients. We also found that the first year after diagnosis was critical for patient survival. On days with elevated PM_2.5 levels, we recommend zimplementing personal protective measures, such as reducing outdoor exposure, using protective masks, and increasing dietary intake of fruits and vegetables [34].

There is a pressing need for continued research and monitoring of air quality and its health impacts, as they are pivotal for developing more precise interventions and policies. Given China’s dense population and severe air pollution issues, particularly in northern regions, the adverse health effects of air pollution warrant increased attention [35]. Several strategies are essential to mitigate pollutants such as BC and OM and consequently reduce lymphoma mortality rates. These include enhancing industrial emissions standards, incorporating cleaner production techniques, and developing advanced filtration systems. Shifting towards renewable energy sources, like solar and wind power, can diminish our reliance on fossil fuels, which are significant sources of these pollutants. Stricter vehicle emission regulations and the promotion of electric and hybrid vehicles are also key to reducing urban air pollution. Raising public awareness about the dangers of these pollutants and advocating sustainable practices are equally important. Furthermore, investing in the development of biodegradable and environmentally friendly materials can significantly reduce the generation of OM. The implementation of these measures not only promotes a healthier environment but also significantly reduces the risk and mortality of lymphoma patients, highlighting the vital link between environmental health and patient outcomes.

Conclusions

Exposure to PM_2.5 during the first year following diagnosis may adversely affect the survival of patients with lymphoma. Our results indicate a potential association between lymphoma prognosis and exposure to OM and black carbon BC, providing a theoretical framework for subsequent mechanistic studies. Furthermore, we recommend that lymphoma patients consider residing in areas with better air quality, which may extend their survival. This study highlights the pivotal role of air quality management in potentially slowing disease progression and improving patient outcomes, underscoring the broader implications of environmental health in cancer care and survival.

Abbreviations

PM_2.5

Particulate matter smaller than 2.5 µm

O₃

Ozone

NK

Natural killer

OR

Odds ratio

CI

Confidence interval

TAP

China Air Pollution Tracking Database

CMAQ

Community Multiscale Air Quality

OM

Organic matter

BC

Black carbon

SD

Standard deviation

HR

Hazard ratio

PFS

Progression-free survival

Authors’ contributions

All authors contributed to the study conception and design. Material preparation, data collection and analysis were performed by CL and DY. The first draft of the manuscript was written by LM, KW and CL. All authors commented on previous versions of the manuscript. All authors read and approved the final manuscript.

Funding

This work was funded by Henan Provincial Medical Science and Technology Research Program Key Soft Science Project (RKX202401004); Clinical Research Special Project of Shanghai Municipal Health Commission (20234Y0007).

Data availability

The datasets analyzed in this study are accessible from the corresponding author upon reasonable request.

Declarations

Ethics approval and consent to participate

The study design was approved by the Medical Ethics Committee of Henan Cancer Hospital (2021-KY-0054-001), and written informed consent was obtained from all participants before initiation of the follow-up process. All procedures involving human participants were conducted in accordance with the ethical standards of the institutional research committee and the principles outlined in the Declaration of Helsinki (1964) and its subsequent amendments.

Competing interests

The authors declare no competing interests.