Ruimin Wang, Shouzhi Yang, Mengfei Wang, Yan Zhou, Xvelian Li, Wei Chen, Wanshan Liu, Yida Huang, Jiao Wu, Jing Cao, Lei Feng, Jingjing Wan, Jiayi Wang, Lin Huang, Kun Qian

Nature Sustainability 2024, 7, 602-615.

Abstract

Over a billion people across the world experience a high rate of missed disease diagnosis, an issue that highlights the need for diagnostic tools showing increased accuracy and affordability. In addition, such tools could be used in ecologically fragile and energy-limited regions, pointing to the need for developing solutions that can maximize health gains under limited resources for enhanced sustainability. Metabolic diagnosis holds promise but faces challenges due to the applicability of biospecimens and limited robustness of analytical tools. Here we present a diagnostic method coupling dried serum spots (DSS) and nanoparticle-enhanced laser desorption/ionization mass spectrometry (NPELDI MS). Our approach allows diagnosis of multiple cancers within minutes at affordable cost, environmental friendliness, serum-equivalent precision and user-friendly protocol. Our assessment shows that the implementation of this tool in less-developed regions could reduce the estimated proportion of undiagnosed cases of colorectal cancer from 84.30% to 29.20%, gastric cancer from 77.57% to 57.22% and pancreatic cancer from 34.56% to 9.30%—an overall reduction in the range of 20.35–55.10%. This work provides insights into delivering more sustainable metabolic diagnosis with maximum health gains.

Introduction and Methods

Metabolic diagnosis using dried spots faces challenges to sustainability in clinical laboratories in terms of the robustness of analytical tools, aiming to deliver accurate results without the need for repeated testing or recalibration, thus reducing the consumption of excessive resources. Mass spectrometry (MS) serves as the mainstream tool for metabolic diagnosis using dried spots, only consuming microlitre volumes of blood. However, tedious chromatographic separation before MS is universally required, considering the complex mixture composition as well as low metabolite abundance in clinical samples. Apart from coupling chromatography with MS, organic matrices (for example, 2,5-dihydroxybenzoic acid, DHB) are critical in laser desorption/ionization (LDI) MS-based metabolic analysis, improving ionization efficiency by converting laser energy to analytes. Limitations of unsatisfactory sensitivity, selectivity and reproducibility still exist when using organic matrices for metabolic analysis.

Therefore, there is an urgent need to tailor inorganic nanoparticles (NPs) for use as matrices in metabolic diagnosis. In particular, nanoparticle-enhanced LDI MS (NPELDI MS) is applied for in vitro metabolic diagnostics, using inorganic NPs as chromatography alternatives to selectively enrich metabolites and yield reproducible data with enhanced sensitivity. However, the adaptation of NPELDI MS to dried spot analysis has not been validated. A further obstacle is the establishment of a complete approach with a standardized operating procedure for diagnostic use. Both challenges must be tackled to meet the needs of sustainability in clinical laboratories towards hierarchical medicine.

Key Results and Conclusions

The proposed approach enables sustainable diagnostic solutions for ecologically fragile and energy-limited regions towards environmental protection. Only room temperature is required for DSS samples in both delivery and storage, which reduces resource consumption and carbon emissions towards low-carbon and climate-positive medicine. With each 1 m2 increase in cooling space, the necessary electrical power and Freon are estimated to increase by ~160 W and ~50 g, respectively. Raw cotton, the primary manufacturing material of DSSs, has a greater degree of sustainability than plastic-based serum cryopreservation tubes in terms of global warming, resource depletion and chemical hazard, with a Higg materials sustainability index of 1.12 and 3.92, respectively. Regarding biosafety, sample drying in DSS preparation can be considered an inactivation process, ruling out the potential risk of infection, with low biosafety concerns. In contrast, serum samples from patients infected with blood-borne infectious agents are highly pathogenic (for example, an estimated viral load of ~12,500 copies ml−1 for HIV in serum62), requiring sensitive virus detection (taking the detection window into account) and additional inactivation/removal technology.

The diagnostic precision of an in vitro diagnostic assay is under strict evaluation according to the guidelines of regulatory authorities to ensure satisfactory analytic performance of the new method compared with the standard. Traditional imaging is difficult with limited medical resources, such as a limited instrument penetration rate (for example, 18 CT scanners per million population in China as compared with 44 in the US), professional physician capacity and high diagnostic cost (for example, US$35 for a CT scan).

Blood-based liquid biopsy facilitates population-based cancer screening programmes and is available in third-party independent medical laboratories as well as routine medical environments. The performance of well-documented blood cancer biomarkers lacks sufficient sensitivity in clinical practice (for example, overall sensitivity of 50–80%/40–60%/63–76% for carcinoembryonic antigen in diagnosing PC63, CC64 and GC65). Serum-based analysis is also constrained in areas with inadequate medical resources owing to the storage conditions needed to guarantee accurate quantitation. Here, robust MS data from a DSS-based approach were merged with a centralized system for chemometric analysis and disease diagnosis beyond typical clinical scenarios. DSS-derived models outperformed the clinically approved biomarkers in diagnosing cancer patients, with serum-equivalent precision (sensitivity of 82–100%) via metabolic diagnosis (Fig. 5e).

The validation of the DSS-based NPELDI MS diagnostic approach for other disease entities requires further research. Development of less expensive and portable MS platforms is urgently required to achieve point-of-care testing in resource-limited areas with high disease burdens.

In summary, we proposed an approach with standardized workflow involving NPELDI MS with paper-based DSS towards sustainable metabolic diagnosis. Based on the optimized workflow, the approach is practical and can achieve a high level of diagnostic accuracy, even when carried out by local health workers in resource-limited clinical settings. This work provides insight into delivering metabolic diagnosis with maximum health gains using available resources.

Fig. 1: NPELDI MS platform.

a, Scanning electron microscopy images of ferric nanoparticles showing nanoscale surface roughness and large-scale uniformity (inset). b, Transmission electron microscopy (TEM) image of ferric nanoparticles and high-resolution TEM image (inset, d refers to interplanar spacing of ferric nanoparticles) showing crystal lattice. c, High-angle annular dark-field image and carbon (indicated with C in yellow), oxygen (O in green) and ferrum (Fe in purple) mappings for nanoparticle–glucose hybrids. d, Digital image of the microarray chip after loading samples and nanoparticles. The distance between two adjacent spots is 5 mm. e, Mass spectra of standard mixture (1 mg ml−1 of Ala, Lys, Arg, glucose (Glc) and sucrose (Suc)) recorded in the m/z range of 100–400 using ferric nanoparticles, CHCA and DHB as matrices in positive ion mode. f, Intensity reproducibility of five standard metabolites using ferric nanoparticles, CHCA and DHB as matrices in positive ion mode at intra-batch level with 9 replicates per sample for each metabolite. g, Bright-field micrographs and corresponding 3D reconstruction images of matrix–analyte co-crystallizations, with ferric nanoparticles (left) compared with CHCA (middle) and DHB (right) serving as matrices. Scale bar, 200 nm (a–c), 1 µm (inset in a), 5 nm (inset in b) and 100 µm (g). Experiments were repeated independently 3 times with similar results (a–c, g).

Fig. 2: Feasibility analysis.

a, Mass spectrum of DSS extraction with NPELDI MS in the m/z range of 100–500. Lac, lactate. b, Typical mass spectra of glucose (m/z: 203) and its 13C-isotope ([13C6] glucose, m/z: 209) as internal standards for quantification, with concentration ratios of analyte:13C-isotope of 0.2, 1, 2, 5 and 10. c, Calibration curve obtained by plotting the intensity ratio of glucose:13C-isotope (defined as I203/I209) as a function of analyte contents. Data are shown as mean ± s.d. (n = 3 independent experiments). d, Linear correlation relationship of glucose in 17 DSSs and serum samples collected from 9 gastric cancer patients and 8 healthy donors. A classic o-toluidine colorimetric method was adopted as the reference method to determine true ‘glucose concentration’ on the x axis. Black dots show the glucose quantitation (quantified by I203/I209) in DSS samples. Blue histogram shows the glucose content in serum, in which the heights of blue bars equal the concentration on the x axis. e, Venn diagram of all MS peaks (defined as S/N > 3) detected across 50 DSS and 50 corresponding serum samples, with peaks over 30% missing ratios in a particular specimen group removed in the subsequent analysis. f, Cosine similarity was analysed between the reference spectrum and the remaining spectra within a specimen group. Vectors for cosine similarity scoring were defined as the reference spectrum (referred to as vector A) and the remaining spectra (referred to as vector B) within a specimen group. g, Metabolite classes detected in DSS and serum samples. h, Principal component analysis score plot of 110 shared metabolite features in 30 DSS and serum samples, with the first three components explaining 24.8%, 7.9% and 4.9% of the total variance.

Fig. 3: Applicability to real-world cases.

a, Schematic workflow for validating the clinical applicability of DSSs, DBSs and serum samples regarding storage conditions (storage atmosphere, temperature and time) and punching locations. b, Metabolite variation in DSSs and paired serum samples under different atmospheres, including air, vacuum and nitrogen atmosphere. Similarity scores were calculated on the basis of the cosine correlation method, selecting samples under vacuum as the reference spectrum. c, Metabolite variation in DSSs and paired serum samples under different temperatures, compared with samples stored at −20 °C. d,e, Long-term stability of DSS (d) and serum (e) samples stored in air at room temperature for up to 144 h, compared with freshly prepared samples. The P values for 24, 48, 72, 96, 120 and 144 h were 0.044, 0.039, 0.013, 0.006, 0.004 and 0.004, respectively. f, Metabolite variation among different locations for three DSSs and paired DBSs, with location 1/2/3/4 referring to central, intermediate and two peripheral regions. Significant differences were calculated by comparing the metabolic profiles at location 2/3/4 with location 1, showing different proportions of statistical differences (P > 0.05, blue; P < 0.05, pink; two-tailed Student’s t-test). g, Glucose quantitation (quantified by I203/I209) in 3 DSSs and paired serum samples under different atmospheres, including air, vacuum and nitrogen atmosphere. h, Glucose quantitation (quantified by I203/I209) in 3 DSSs and paired serum samples under different temperatures. i, Glucose quantitation (quantified by I203/I209) in DSSs and paired DBSs, with location 1/2/4 referring to central, intermediate and peripheral regions. P values for DSS: 0.060 (L1 vs L2), 0.132 (L1 vs L4) and 0.210 (L2 vs L4); for DBS: 0.002 (L1 vs L2), 1.197 × 10−5 (L1 vs L4) and 1.071 × 10−5 (L2 vs L4). *P < 0.05; NS, no significant difference (P > 0.05); two-tailed Student’s t-test. Data are shown as mean ± s.d. (n = 3 independent experiments).

Fig. 4: Validation via cancers diagnosis.

a, Standardized workflow of DSS-based NPELDI MS in cancer screening, including sample collection, serum isolation, DSS preparation, sample loading, metabolic detection and chemometric analysis. b, Distribution of c.v.s of intensities for the apparent molecular peaks in four randomly selected DSS samples (including 2 patients and 2 healthy donors (HD)) at an intra-batch level (n = 9 independent experiments). Data are presented as median ± s.d. c,d, Supervised sparse partial least squares-discriminant analysis of metabolic profiles extracted from serum (c) and DSS (d), showing similar separation between HDs (blue) and CC (pink). e, ROC curves for the classifier designed to distinguish between HDs and CC, by consuming trace DSS (solid lines) and serum (dashed lines) samples. Blue refers to ROC curves obtained by 20-fold cross-validation for the training set; pink refers to ROC curves obtained from a single blind test for the test set; purple refers to ROC curves obtained from the training set via a biomarker panel. f, Scatterplot of feature selection by comparing the fold change between HD and CC in DSS and serum samples. g, Biomarker panel in CC. Gal, galactitol; Mal, malic acid; Gly, glycerol.

Fig. 5: Economic assessment.

a, Sustainability of DSS-based NPELDI MS diagnostic workflow compared with serum in terms of time and economic cost, environmental friendliness and serum-equivalent precision. b, Detection performances of NPELDI MS, nuclear magnetic resonance (NMR) and liquid chromatography (LC)–MS. c, DSS facilitated clinical sample transport and occupied a smaller area (DSS: a 13-mm-diameter sheet delivering up to 40 µl of serum; serum: a ⌀10 mm × 50 mm cryopreservation tube delivering up to 40 µl of serum). d, Sustainability of the filtration cards for DSS preparation (made of raw cotton) compared with cryopreservation tubes for serum preparation (usually made of polypropylene (PP), polyethylene (PE) and polycarbonate (PCB)), in terms of global warming, resource depletion and chemical hazard, scored by Higg materials sustainability index (MSI) (https://higg.com/) according to the Sustainable Apparel Coalition. e, Performance comparison of computer-aided and expert-based diagnostic routes for cancer screening. Scale bar, 2.5 cm (c).

https://www.nature.com/articles/s41893-024-01323-9