The research team, led by Qian Chen and Yuhong Xie of Tianjin University of Technology together with colleagues at the Tianjin Key Laboratory of Hazardous Waste Safety Disposal and Recycling Technology, collected composite effluent samples from the final discharge outlets of all ten enterprises. Rather than hunting for a predefined list of target pollutants, they deployed liquid chromatography coupled to high-resolution mass spectrometry, or LC–HRMS, in a non-target screening mode. This analytical strategy does not assume in advance what is present. Instead, the instrument records the exact masses and fragmentation patterns of thousands of molecular features in each sample, and computational pipelines then attempt to annotate those features by matching them against spectral libraries and chemical databases. In total, the team detected 1,862 distinct features and confidently annotated 347 compounds spanning 20 different chemical classes, a snapshot of the extraordinary chemical diversity that flows out of pharmaceutical production lines even after on-site treatment.

The central analytical challenge was to separate the noise from the signal: with hundreds of compounds in play, which ones are genuinely shared across the industry rather than idiosyncratic to one factory? The researchers applied a battery of complementary statistical tools. Principal component analysis condensed the enormous data matrix into its dominant axes of variation, revealing that the relative abundance profiles of the shared contaminants varied mainly along a single continuous gradient that ran across all ten enterprises. Exploratory Spearman co-variation analysis then examined whether groups of compounds rose and fell in synchrony across samples, a pattern that can hint at shared sources, similar production processes or common treatment behaviors. From this combined evidence, and using MS/MS spectral matching together with relative peak-area criteria, the team narrowed the field to fourteen contaminants that were detected in every single one of the ten enterprises. These fourteen molecules represent the closest thing the industry currently has to a universal effluent signature.

Detection alone, however, does not make a pollutant dangerous, and this is where the study makes its most consequential methodological contribution. The authors evaluated each of the fourteen shared contaminants using the BIOWIN suite of biodegradability prediction models, which estimate how readily a molecule will be broken down by microorganisms in treatment systems and receiving waters. Eight of the compounds were classified as readily biodegradable, meaning that conventional biological treatment has a reasonable chance of eliminating them before they reach rivers or streams. Six, however, were flagged as not readily biodegradable, and these persistent molecules are the ones most likely to survive treatment, accumulate in aquatic environments and exert chronic pressure on ecosystems. Persistence is only half of the risk equation, though; the other half is toxicity, and here the researchers turned to the concept of the predicted no-effect concentration, or PNEC, the concentration below which a chemical is not expected to cause adverse effects.

The PNEC evidence evaluation added a crucial layer of nuance. For some contaminants, robust ecotoxicological data exist and a reliable no-effect threshold can be derived; for others, the underlying evidence is sparse, uncertain or extrapolated from very few studies, which means any risk calculation built on them is only as strong as its weakest input. Rather than treating all PNEC values as equally trustworthy, the team explicitly graded the quality of the evidence behind each threshold. They then combined four independent dimensions — predicted biodegradability, PNEC evidence quality, relative peak-area signal intensity from the LC–HRMS data, and the confidence level of each compound annotation, following the widely used Schymanski confidence scale — into a screening-level prioritization framework. The output was not a simple ranking by abundance or detection frequency but a multidimensional assessment of which compounds warrant immediate, targeted follow-up.

The headline result is striking. Two compounds, norethindrone and norgestrel, emerged as high-priority candidates for screening-level monitoring. Both are synthetic progestins, steroid hormones used in hormonal contraceptives and hormone therapy, and both belong to the not-readily-biodegradable group. Their significance is amplified by biology: synthetic progestins are among the most potent endocrine-active substances known, capable of disrupting the reproductive systems of fish and other aquatic organisms at concentrations in the nanogram-per-liter range. That such compounds appear in the final effluent of every one of ten different pharmaceutical enterprises suggests a systemic exposure pathway rather than an isolated accident, and it raises pointed questions about whether current treatment trains at manufacturing facilities are designed to remove steroid residues at all. A third compound, the surfactant-like substance C12-AE1S, was designated a tentative tracking candidate, pending structural confirmation and evaluation of how it behaves during wastewater treatment.

Perhaps the most important message of the study is a negative one: monitoring priorities should not be set by detection frequency or signal intensity alone. A compound that appears in every sample and produces a large chromatographic peak is not necessarily the one that poses the greatest ecological threat. A less abundant but persistent, highly toxic and poorly characterized molecule may deserve attention first. By the same token, a readily biodegradable compound with strong PNEC evidence may already be adequately controlled by existing treatment processes. This reframing matters for regulators, who face limited analytical budgets and cannot quantify every one of the nearly two thousand features that modern instruments can detect. A defensible prioritization framework, one that weighs persistence, toxicity evidence, exposure signal and identification confidence together, offers a practical route from the overwhelming complexity of non-target screening data to a short, actionable list of compounds for targeted quantification.

The study also arrives at a moment of intensifying global scrutiny of pharmaceutical manufacturing effluent. The World Health Organization issued guidance in 2024 on wastewater and solid waste management for antibiotic manufacturing, reflecting growing concern that production-scale discharges contribute to the spread of antimicrobial resistance and to environmental contamination in regions that host concentrated pharmaceutical production. Prioritization exercises of the kind performed here complement such policy efforts by supplying the empirical groundwork: before discharge limits can be set or treatment technologies mandated, regulators need to know which specific chemicals are actually present, how consistently they occur, and what concentrations would be harmful. The screening-level approach demonstrated across these ten enterprises could, in principle, be extended to other industrial sectors and other regions, building a comparative picture of shared industrial contaminants worldwide.

There are, of course, important caveats. Non-target screening identifies compounds with varying degrees of confidence, and the authors themselves flag that C12-AE1S requires structural confirmation before any firm conclusions can be drawn about its behavior or risk. Peak-area-based relative signals are not the same as calibrated concentrations, so the framework is explicitly designed as a first-pass triage: its purpose is to select candidates for follow-up targeted quantification, source investigation and treatment verification, not to deliver final risk numbers. The raw LC–HRMS datasets and enterprise-level sample information are not publicly available because they contain confidential industrial information, although the processed data and Python scripts underlying the analyses are available from the corresponding author on reasonable request. Those constraints are the price of persuading companies to open their discharge outlets to scientific scrutiny at all.

Even with those limitations, the study offers a template for how environmental chemistry can keep pace with industrial complexity. The combination of high-resolution non-target screening, multivariate statistics, biodegradability prediction and evidence-graded risk thresholds turns a flood of anonymous molecular features into a ranked, defensible monitoring agenda. For the pharmaceutical industry, the findings are a reminder that the same molecules manufactured in different factories, under different processes and with different treatment systems, end up sharing a common chemical signature in the water leaving the plant — and that two of the most consistent members of that signature are potent synthetic hormones. For regulators and treatment engineers, the message is that the next generation of pharmaceutical wastewater monitoring should be driven by risk-informed prioritization rather than by the size of a chromatographic peak, and that persistent endocrine-active residues deserve a place at the very top of the list.

Subject of Research: Occurrence and prioritization of shared organic contaminants in pharmaceutical manufacturing wastewater

Article Title: Cross-enterprise occurrence patterns and monitoring priorities of shared organic contaminants in pharmaceutical wastewater

Article References: Chen, Q., Li, M., Shao, M., Lei, J., Fei, X., Zheng, W., & Xie, Y. (2026). Cross-enterprise occurrence patterns and monitoring priorities of shared organic contaminants in pharmaceutical wastewater. Environmental Science and Pollution Research. https://doi.org/10.1007/s11356-026-38295-6

Image Credits: AI Generated

DOI: 10.1007/s11356-026-38295-6

Keywords: pharmaceutical wastewater, non-target screening, LC-HRMS, norethindrone, norgestrel, synthetic progestins, biodegradability, predicted no-effect concentration, monitoring prioritization, emerging contaminants, endocrine disruption, China