Introduction

Paraquat (PQ) is a bipyridyl, rapidly acting, nonselective herbicide widelyused in the developing world1,2. It is a highly toxic compoundto human beings, with no known antidote3,4. Intentional oraccidental acute PQ poisoning is unfortunately common and many fatal caseshave been reported in China5,6. Poisoning cases are most oftena result of suicide attempts via oral self-administration7.

PQ is largely secreted unchanged in urine within the first 24 hoursof ingestion. To date, the most widely accepted mechanism underlying PQ intoxicationis oxidative stress. Biotransformation of PQ in cells results in the generationof superoxide anions and subsequently other free radicals, resulting in cellularinjury such as lipid peroxidation and mitochondrial dysfunction, triggeringan inflammatory response8,9,10. Uric acid (UA) is the majorend product of purine metabolism and is formed from hypoxanthine and xanthineby the rate-limiting enzymatic action of xanthine oxidoreductase (XO)11,12. It has been reported that PQ treatment increases XO activityand stimulates hypoxanthine-dependent superoxide production in the cytosolof rat lungs13,14. Our previous study indicated increasedXO activity accompanied by lipid peroxidation and reduced total antioxidantcapacity in subjects with acute PQ poisoning15.

Paraquat poisoning is characterized by multiple organ function failure,mainly involving the lung, kidney, heart, liver and nervous system16.Yu et al. observed that treatment with UA could protect neurons against excitotoxicand metabolic insults involving suppression of oxyradical accumulation, stabilizationof calcium homeostasis and preservation of mitochondrial function17.Conversely, Sakai and colleagues reported that the use of allopurinol as adrug to block the production of UA can alleviate intracellular free radicalproduction and reduce PQ cytotoxicity in cultured bovine pulmonary arteryendothelial cells18. A recent study also found that basal levelsof UA in mice do not appreciably protect against oxidative damage and neurotoxicityin the PQ model of Parkinson's disease19. To date, anyassociation between UA and PQ exposure remains uncertain in the literature.Our hypothesis is that elevated UA is a valuable prognostic factor for adverseoutcomes after PQ poisoning. Because of a lack of specific antidotes, theoverall mortality from acute PQ poisoning is substantially high20.This raises the need to develop a valuable predictor for prognosis to guidefuture therapeutic intervention. Further clarification of serum UA levelsin patients with PQ poisoning may thus have significant clinical implicationsby setting a framework for modulating serum UA levels.

Methods

Subjects

This study was a retrospective observational cohort study of patients presentingto the emergency room (ER) of The First Affiliated Hospital, College of Medicine,Zhejiang University between January 2009 and June 2014. We enrolled 221 patientswith an oral intake of paraquat from 2 to 30 mL. Patients who met thefollowing criteria were excluded: those with a history of gout (n = 8), diabetesmellitus (n = 3), hypertension (n = 2), renal failure (n = 2) and malignancy(n = 1). The remaining 205 patients [median age: 33.0 years (range: 14–82years); female patients: 110; male patients: 95] were included in thepresent study. Because UA concentrations differ significantly by gender, patientswere categorized into gender-specific tertiles based on their UA level: tertile1, UA < 329 µmol/L for men and UA < 237 µmol/Lfor women; tertile 2, 329–431 µmol/L for men and UA 237–334 µmol/Lfor women; tertile 3, UA > 431 µmol/L for men and UA >334 µmol/L for women. Informed consent was obtained from allthe subjects.

Sample collection and biochemical analyses

A peripheral venous blood sample (6 mL) was collected from eachpatient within the first 24 hours after admission to the ER. Bloodsamples were used to analyze the hematological index and biochemical values.Laboratory parameters measured included: white blood cells (WBC), platelets,hemoglobin, red blood cell distribution width (RDW), neutrophil-lymphocyteratio (NLR), prothrombin time, total protein, albumin, alanine aminotransferase(ALT), aspartate aminotransferase (AST), lactate dehydrogenase (LDH), creatinekinase (CK), creatinine, potassium, pH, PaCO2 and PaO2.All biochemical analyses were conducted using a Hitachi 7600 Clinical Analyzer(Hitachi, Tokyo, Japan), Sysmex CA-7000 System (Sysmex, Kobe, Japan), andSysmex XE-2100 Automated Analyzer (Sysmex) using standard methods.

Data and statistical analysis

Statistical analyses were performed using SPSS, version 16 (SPSS, Chicago,IL, USA). Data are presented as the mean ± standard deviation whendata were found to be normally distributed or as the median if the distributionwas skewed. The differences among multiple groups or between two groups wereassessed using a one-way analysis of variance (ANOVA) and the Kruskal–WallisH test or Mann–Whitney U test, if appropriate. Differences by genderand in 30-day mortality among the groups were compared using a Chi-squaredtest. The area under the ROC curve was used to discriminate UA levels withrespect to 30-day mortality. Univariate and multivariate Cox regression analysesto determine predictors of 30-day mortality were presented as hazard ratioswith a 95% confidence interval. Variables that showed a P value <0.05 in the univariate analysis were included in the multivariate analysis.All statistical tests were two-tailed. P < 0.05 was considered significantlydifferent.

Ethics statement

This study was approved by the ethics committee of The First AffiliatedHospital, College of Medicine, Zhejiang University and was conducted in accordancewith the Declaration of Helsinki.

Results

Patient characteristics

We divided patients with PQ poisoning into three groups [Tertile 1(lowest), Tertile 2 and Tertile 3 (highest)] according to the tertileof their serum UA levels. As shown in Table 1, wefound that time from PQ ingestion to ER admission was significantly differentbetween the lower two tertiles and the highest tertile. Across increasingserum UA tertiles, WBC, NLR, ALT, AST, LDH, CK and creatinine levels weregradually increased, while potassium, arterial pH and PaCO2 graduallydecreased.

Table 1 Baseline laboratory characteristicsaccording to UA tertile
Full size table

Association between serum UA level and biochemical variables

Serum UA levels were significantly and positively correlated with inflammatoryindexes [WBC (r = 0.196, P < 0.05), LDH (r = 0.178, P <0.05) and NLR (r = 0.173, P < 0.05)] and markers of multi-organdamage [ALT (r = 0.171, P < 0.05), AST (r = 0.192, P <0.05), CK (r = 0.186, P < 0.05) and creatinine (r = 0.398, P <0.05)]. Serum UA levels were also negatively correlated with parametersof arterial blood gases [pH (r = −0.161, P < 0.05) andPaCO2 (r = −0.171, P < 0.05)] and potassium(r = −0.182, P < 0.05).

Comparison of the serum UA level between the survival group andnon-survival group

As shown in Figure 1, the serum UA level in the non-survivalgroup was significantly higher than in those who survived (373.3 ±51.2 µmol/L vs 321.6 ± 60.1 µmol/L, P <0.001).

Figure 1
figure 1

Comparison of serum UA level between the survival group and non-survivalgroup.

Data are mean ± SD. *P < 0.05 compared with thesurvival group.

Full size image

Association of UA level with 30-day mortality rate

Sixty-six patients died during the first 30 days after admission to theER, corresponding to a 32.2% cumulative incidence of mortality. To get a deeperunderstanding of the relationship between UA level and PQ poisoning, the cumulative30-day mortality of patients was calculated by dividing the number of fatalitiesby the number of subjects in each UA tertile (Table 2).The 30-day mortality rate tended to increase as the UA level increased. Comparedwith only 17.1% in tertile 1, the 30-day mortality rates for the subjectsin tertile 2 and tertile 3 were 32.4% and 47.8%, respectively.

Table 2 30-day mortalityaccording to serum UA tertile
Full size table

Optimal UA cut-off value for predicting 30-day mortality

The cut-off point for UA to predict 30-day mortality in female patientswas 284 µmol/L and the area under the receiver operating characteristic(ROC) curve was 0.752 (95% confidence interval, 0.655–0.849, P <0.001) (Figure 2). When the UA was >284 µmol/Lin female patients, the sensitivity was 82.9% and the specificity was 68.5%.The cut-off point for UA in male patients was 352 µmol/L andthe area under the ROC curve was 0.732 (95% confidence interval, 0.637–0.829, P <0.001). The sensitivity was 79.2% and the specificity was 60.5%.

Figure 2
figure 2

Cut-off point for UA for predicting 30-day mortality from paraquatpoisoning.

Full size image

Risk factor analysis for 30-day mortality

As shown in Table 3, univariate analysis showeda significant association of WBC, creatinine, LDH, CK and UA with 30-daymortality. In the multivariate Cox proportional hazards regression analyses,WBC and UA were independent prognostic factors.

Table 3 Cox regression analysisof risk factors for 30-day mortality in patients with PQ poisoning
Full size table

Discussion

Various studies have shown that PQ primarily exerts its toxic effects throughthe redox cycle, which produces oxygen free radicals, leading to oxidativedamage and eventual cell death21,22,23. Many studies havesought to evaluate outcome indicators of PQ intoxication, but there is stillno consensus on a practical indicator. To date, no prognostic models havebeen prospectively validated because of issues in their development such assmall sample size, differences in the degree of severity and complicatedexclusion criteria24. The urine dithionite test is a simpleindex for clinical diagnosis and prediction of prognosis of PQ poisoning;however, false results limit its usefulness25. Our study isone of the few performed to date addressing serum UA levels and laboratoryparameters in a clinical context with relatively large sample size. Our resultsconfirmed our hypothesis: 30-day mortality increased with progressively higherbaseline serum UA level and increased UA level was found to be an independentprognostic factor in patients with acute PQ poisoning. Recently, there hasbeen growing interest in this parameter because increased UA is a strong independentbiomarker of adverse outcomes in many diseases and conditions linked to increasedoxidative stress and inflammation26,27,28. UA is formed viathe purine degradation pathway by the action of XO and is excreted by thekidney into the urine29,30. Serum UA levels are rigorouslycontrolled by a balance between UA synthesis and excretion. One of the plausibleexplanations for an increased UA level in human PQ poisoning is upregulationof XO activity. Studies have confirmed that PQ treatment significantly inducesXO activity and increases hypoxanthine-dependent superoxide production13,14,18, which would then result in increased UA as it is theother main product of XO activity. Similarly, we previously reported significantlyhigher serum XO activity in a PQ poisoning group vs healthy controls15. In addition to the increased generation of UA, another possiblemechanism for a rise in serum UA is reduced excretion. The kidney is consideredthe main excretory organ for PQ in humans, eliminating it in the form of urine31. Robust epidemiological data have shown PQ-induced kidney injurysuch as acute tubular necrosis in the proximal tubule, interstitial inflammation,and impaired glomerular filtration rate32,33. Renal dysfunctionleads, in turn, to decreased serum UA clearance.

These results provide novel evidence for a strong association between UAand the mechanism of PQ toxicity. Several mechanisms could explain the significantrelationships between serum UA and mortality as a result of PQ poisoning.The association between hyperuricemia and increased risk of metabolic syndromeand cardiovascular disease has been evaluated and hyperuricemia has beenshown to be a strong and independent predictor of all-cause mortality in community-basedstudies34,35. Over the past few years, epidemiological studieshave repeatedly demonstrated that UA acts as a strong oxidant and an elevatedserum UA level may stimulate oxidative stress and endothelial dysfunctionin several pathological states36,37. UA also triggers an inflammatoryresponse by stimulating the production of proinflammatory cytokines38.Increased systemic oxidative stress and inflammation play a crucial role inthe development of PQ-induced injury both in animal experiments and clinicalstudies39,40,41. The correlation between UA and inflammatoryindices and markers of organ damage in the present study also support thatUA may be a significant risk factor for the development of PQ poisoning. Furtherinvestigation regarding the association between UA and PQ poisoning will expandour understanding of this toxicity.

This study has some limitations. First, we can only propose a role forUA in the etiology of PQ poisoning based on a snapshot of the circulatingUA state. This observational study was unable to definitively comment on causalityor the temporal association between high serum UA and PQ poisoning. The secondlimitation is that blood PQ levels were not assessed and PQ poisoning caseswere included on the basis of a history of oral PQ ingestion. Although, theprognostic value of plasma PQ has been previously documented in subjects withacute PQ poisoning42,43. Unfortunately, this assay is notcommonly available in the ER owing to limited medical facilities44.In addition, traditional methods for detecting PQ levels are time consumingand most ER physicians do not rely on the results of plasma PQ measurementsfor emergency management decisions45.

In summary, our results demonstrated that measurement of serum UA may bea simple and practical index for assessing the outcome of PQ poisoning. Basedon this finding, interventions that aim to decrease UA levels may have beneficialeffects in treating patients with acute PQ intoxication. Nevertheless, furtherstudies verifying the precise role of UA are needed to eventually guide developmentof clinical intervention strategies for PQ poisoning.