In two distinct time periods, we conducted the study in eastern India’s malaria-endemic Kolkata and Purulia. The first phase samples were gathered between April 2008 and November 2010 before and around the time that India’s new drug policy went into effect. After 6–8 years of the new national drug policy, we also collected the samples between March 2015 and December 2017 for the second phase. Patients who experienced a fever with headache, shaking, or vomiting within the previous 24 h and those who had not taken any antimalarial medicine during that time were submitted for a malaria test. We used EDTA-coated vacutainer tubes to collect 3 ml of intravenous blood from 4,652 suspected patients in accordance with the Helsinki procedure.

P. falciparum mono-infection determined by microscopic analysis of Giemsa-stained thin and thick blood smears, patients with a parasite density of 1,000–10,000 asexual parasites/l blood, and patients without a history of self-medication with antimalarials were the inclusion criteria for our study (Bloland et al., 2003). Through the use of an allelic family-specific nested PCR (MAD20 and K1 for pfmsp-1 and 3D7 Africa and FC27 for pfmsp-2), monoclonal P. falciparum infections were further evaluated (Das et al., 2014). Pregnant women, newborns, people with non-falciparum malaria, people with severe complicated malaria, and people with mixed infections (P. falciparum + P. vivax) were excluded from the study (Bloland et al., 2003). The study also eliminated any individuals who had used antibiotics or antimalarial medications prior to the diagnosis of P. falciparum infection. We obtained each patient’s or the patient’s legal guardian’s written informed permission. The Vidyasagar University Human Ethical-Committee gave its full approval to the experimental design and protocols (VU/HEC 08-012).

P. falciparum– positive patients received standard SP doses under the supervision of a medical officer in accordance with the national guideline, with adults receiving a single oral dose of three tablets containing 1000 mg of SDX and 50 mg of PYR, whereas children received a single dose of 20 mg/kg of SDX and 1 mg/kg of PYR (Guidelines for Diagnosis and Treatment of Malaria in India, 2009). Within their expiration dates, we used the quality-assured medications duly provided by the Ministry of Health and Family Welfare, Government of India (IPCA, LTD). Giemsa-stained thin blood smears from each patient were prepared in triplicate at intervals of 12 h until the parasite became undetectable. Two skilled microscopists examined blood smears. If any discrepancies in the results were detected, a senior microscopist was consulted for confirmation. On days 1, 2, 3, 4, 7, 14, 21, and 28, treatment follow-ups were conducted to assess the patient’s clinical status and assure their safety (Bloland et al., 2003). Additionally, if malaria symptoms returned in the intervals between the scheduled visits, unscheduled follow-ups were conducted. By analyzing the msp1, msp2, and glurp genes using family-specific nested PCR, it was possible to distinguish between reinfection and recrudescence (Das et al., 2013b).

According to WHO guidelines, we divided the therapeutic efficacy into three categories: early treatment failure (ETF), late treatment failure (LTF), and adequate clinical parasitological response (ACPR) (WHO, 2013). Patients who did not respond to SP were given a regular dose of artemether-lumefantrine (six tablets, each containing 40 mg AM and 240 mg LF) as a rescue medication and were then monitored for the following 28 days.

After adapting clinical isolates to in vitro growth using our standard protocol, the phenotypic parameters of in vitro SP-sensitivity were assessed (Das et al., 2013a). Following our standard laboratory protocol, P. falciparum clinical isolates were grown and maintained in culture (Das et al., 2014). The culture plates were incubated at 37°C with a 95% relative humidity environment and 5% CO₂. After 48 h of incubation with SDX and PYR using our routine laboratory procedures, we measured the 50% inhibitory concentrations (IC50) of SP using a hypoxanthine incorporation assay (Das et al., 2014). We used nonlinear probit/logit regression analysis to express the values (Basco and Ringwald, 2000). Stock and dilutions of PYR and SDX were made, respectively, with ethanol and dimethyl sulphoxide (DMSO, 0.1%). The final concentrations ranged from 50 to 25600 nM for both PYR and SDX. We studied each concentration in duplicate. However, we used the PYR-sensitive and SDX-resistant-3D7 strains as control strains and adhered to the established threshold values of in vitro SP sensitivity as previously described (Basco and Ringwald, 2000; Das et al., 2012a).

We isolated the parasite DNA from parasitized erythrocytes as mentioned earlier (Basco and Ringwald, 2000). Furthermore, we used the P. falciparum wild-sequence (AL844509.2) available in GenBank to create the primers for the pfdhfr and pfdhps genes using the primer design software Primer3. Using nested PCR, followed by restriction fragment length polymorphism and sequence analysis, we evaluated the single nucleotide polymorphisms of the pfdhfr and pfdhps genes (Das et al., 2013a). We used an ABI PRISM big-dye terminator cycle sequencing kit and a 3730-xl genetic analyzer to sequence the pfdhfr and pfdhps amplicons in order to confirm the mutation by reading both the forward and reverse strands. The sequence was translated using the Protein Analysis System proteomic service (http://www.expasy.org), and the translated sequences were aligned using ClustalW2 (http://www.ebi.ac.uk/clustalw).

We have discovered a novel pfdhfr double mutant haplotype. Using the crystal structure of the DHFR protein from Plasmodium falciparum (PDB id: 6A2K) as a template, the double mutant DHFR protein’s 3D structure was modeled in Phyre-2 (Kelley et al., 2015). PyMOL was used to superimpose the wild-type and mutant-type proteins to test the model’s accuracy (The PyMOL Molecular Graphics System, Version 1.8 Schrödinger, LLC, 2016; Badawi et al., 2016).

In vitro IC₅₀ values for PYR and SDX coupled with parasitemia were provided as mean SD data. Fisher’s exact test and regression analysis were used to investigate the relationship between SP treatment efficacies and various genotypes. Through the use of the Fisher’s exact test and the Mann–Whitney U test, we evaluated the associations between the IC₅₀s of SP and the dhfr and dhps genotypes. The Kruskal–Wallis test was also used to compare data among more than two groups. The total cure-rate was evaluated using the Kaplan–Meier estimate. The 95% confidence intervals were calculated using the Clopper-Pearson formula. Multivariate logistic regression analysis was used to evaluate the connections between pfdhfr, pfdhps mutations, and in vitro SP resistance. When p < 0.05, a statistical difference was deemed significant. With the help of GraphPad software 3.0. In Stat and Origin 6.1, we conducted the statistical analysis.

In this investigation, we have paid particular attention to the duration-dependent antifolate susceptibility of P. falciparum isolates. Eight hundred three (17.2%) of the 4,652 suspected cases of malaria had finished the 28-day treatment follow-up. These individuals had uncomplicated P. falciparum monoclonal infections and had a mean age of 27 years (95% CI: 8–57). Of them, 335 patients were from Purulia, whereas the rest 468 patients were from Kolkata ( Supplementary Table S1 ). A total of 761 patients (16.4%) who were P. vivax–positive were eliminated. Due to combined P. falciparum + P. vivax infection, an additional 181 individuals (3.9%) were removed. While men (57.7%) were found to be more affected in Kolkata, women (53.4%) were found to be more impacted in Purulia. Table 1 lists the unique traits of participants upon enrolment in the two study locations. Details of patient selection were presented in Figure 1 .

In 2008–2009, 318 P. falciparum– infected patients had finished the 28-day SP therapy follow-up. A total of 221 of them were from Kolkata, whereas 97 were from Purulia. A summary of the treatment effectiveness following SP therapy is provided in Supplementary Table S2 .

In addition, 485 additional P. falciparum infections were treated with the artesunate-SP combination in accordance with the national guidelines between 2010 and 2015–2017; however, we were unable to disclose the clinical outcome data due to a lack of sufficient authorization. The genomic characterization of the antifolate resistant genes in those isolates was presented together with each isolate’s unique phenotypic traits of in vitro SP susceptibility. However, 205 patients (64.5%) out of 318 P. falciparum– infected patients who got SP therapy experienced a successful outcome (ACPR).

More specifically, SP was effective in treating 61.7% of the isolates from Kolkata and 72.1% of the isolates from Purulia. Early treatment failure (ETF) was found in 25.8% of patients based on day 3 positive parasite (parasitemia >1% that of day 0), associated with continuation of fever on day 3, whereas 9.8% of patients were diagnosed as late treatment failure (LTF). Among the study participants, no deaths have been documented. Following the SP treatment, parasite recurrence was seen in three patients on day 4, eight patients on day 7, five patients on day 14, 19 more patients on day 21, and nine more patients on day 28.

All 44 cases of apparent parasite reappearance were subjected to an analysis of merozoite surface proteins 1, 2, and glurp, of which 31 (70.4%) isolates were identified as cases of late SP treatment failure or true recrudescence ( Supplementary Table S2 ). Treatment failure for SP in Purulia (8.9% ETF + 8.9% LTF) was recorded in 17.8% of patients in 2008 and increased to 36.5% (19.2% ETF + 17.3% LTF) in 2009. In Kolkata, treatment failure for SP (6.7% ETF + 8.9% LTF) was only recorded in 15.6% of patients in 2008 and increased sharply to 44.9% (36.9% ETF + 7.9% LTF) in 2009. In contrast to Purulia, Kolkata had a higher prevalence of ETF following SP therapy ( Supplementary Table S2 ). The overall cure rate following SP therapy was reported to be 61.1% in Kolkata (Kaplan–Meier estimate; 95% CI 54.29–67.48) and 72.16% in Purulia (61.98–80.55). All the treatment failure cases were successfully treated after AMLF therapy.

The phenotypic characteristics of in vitro sensitivity for PYR and SDX yielded interpretable results in 795 and 787 isolates, respectively. In Kolkata, the percentage of isolates that were PYR-sensitive in 2008 was 24.4% (mean IC50 = 71.4 nM, 95% CI, 49–91), but it sharply declined to 6.4% by 2017. In Purulia, PYR sensitivity was detected in 46.7% of isolates in 2008 (mean IC50 = 68.8 nM, 95% CI, 51–88 nM), but it dropped to 6.9% in 2017 (mean IC50 = 72 nM, 95% CI, 55–95). In comparison with 2008, the proportion of intermediate PYR resistance significantly decreased at both research locations in 2017. In Purulia, intermediate PYR resistance was just 6.9% in 2017, whereas in Kolkata, intermediate PYR resistance decreased from 28.9% in 2008 to 10.6% in 2017.

In contrast, in Kolkata, despite no usage of SP since 2009, 83% of isolates were found to be extremely resistant to PYR in 2017 (mean IC50 = 9770.65 nM; 95% CI, 7150–12380), up from 46.7% in 2008 (mean IC50 = 5606 nM; 95% CI, 3155–8168). Additionally, the percentage of PYR-resistance in Purulia gradually grew from only 26.7% in 2008 to as high as 86.2% (mean IC50 = 9120.5 nM; 95% CI, 6075–12065) in 2017 ( Figure 2A ).

In Kolkata, the majority of SDX-sensitive parasites were found to be predominant (53.3%) in 2008 but significantly declined to 6.4% in 2013. In 2008, only 17% of isolates were reported to be SDX resistant (mean IC50 = 6015.35 nM; 95% CI, 3988–8050); however, this number rose dramatically to 85.1% in 2017 (mean IC50 = 10220.5 nM; 95% CI, 7800–12500). In Purulia, 55.6% of isolates were found to be SDX-sensitive in 2008, but this figure had dropped to 12.1% in 2017 (mean IC50 = 351.5 nM, 95% CI, 155–560). SDX resistance was gradually increased to 79.3% in 2017 (mean IC50 = 9670 nM; 95% CI, 6980–12390 nM) from only 15.6% in 2008 ( Figure 2B ). Pyrimethamine and sulfadoxine in vitro responses were highly correlated (r ² = 0.9965, p < 0.019). On the other hand, intermediate SDX resistance in Kolkata fell to 8.5% in 2017 from 29.7% in 2008, whereas intermediate PYR resistance in Purulia was around 8.6% in 2017.

In 2008, 71.1% of Kolkata isolates had the pfdhfr-S108N mutation, which was slightly reduced to 69.3% in 2009. In 2009, 45.8% of Kolkata isolates and 26.9% of Purulia isolates encoded novel mutant isoleucine at codon108 (108I) instead of wild serine or mutant asparagine or threonine (KP256234, KP256235, KP256236, and KP256237) ( Figures 3A–C ). This novel dhfr-108I allele was always observed with dhfr-51I mutation and was only discovered after super-cyclone Aila for 6–8 months. After that, we never saw this haplotype again. The superposition of a novel double mutant (108I + 51I) allele with the wild allele revealed structural uniformity except at the 51 and 108 codons ( Figure 3D ); this resulted in potential conformational changes in the PfDHFR protein, which reduced the binding energy during interaction with pyrimethamine ( Figures 3E, F ), resulting in resistance to pyrimethamine. In 2009, the proportion of Purulia isolates with the most vulnerable S108N or novel S108I mutation increased from 51.1% to 86.5%. In 2008, the N51I mutation predominated in both Kolkata (71.1%) and Purulia (66.7%). In 2009, the N51I mutation was reduced to 67.1% in Kolkata but increased to 80.7 percent in Purulia. In 2008–2009, no mutations were found in the pfdhfr gene at codon-A16V and I164L ( Figure 2C ).

In Kolkata, the core pfdhps-A437G mutation was found in 51.1% of isolates in 2008 and 56.8% of isolates in 2009, respectively. From 2008 to 2009, Purulia saw a substantial rise in the percentage of isolates with the pfdhps-A437G mutation, going from 55.6 to 80.8%. In Kolkata, the prevalence of the S436A mutation significantly rose from 2008 (37.8%) to 2009 (61.4%). In Purulia, the percentage of the mutant S436A-allele was similar in 2008 (55.6%) and 2009 (61.5%). Only 13.5% of the isolates in Purulia had the A613T mutation, compared with a dramatic increase of 35.2% in Kolkata in 2009 ( Figure 2D ).

In Kolkata pfdhfr-A I C N I haplotype was prevalent (31.1%) followed by A IRN I (26.7%) and wild ANCSI (15.6%) allele in 2008, whereas in Purulia pfdhfr-A I C N I (40%) was most frequently found haplotype after A I CSI (26.7%) and ANCSI (22.2%) allele. In 2009, novel pfdhfr-A I C I I haplotype (45.5%) was predominant followed by ANCSI (30.7%) and A I C N I allele (18.2%) in Kolkata, while in Purulia, A I C N I (42.3%) haplotype was commonly found after A I C I I (26.9%) and ANCSI (13.5%) allele ( Figure 2E ). In Kolkata, wild pfdhps-SAKAA haplotype (28.9%) was prevalent after S G KAA (22.2%) and A AKAA (15.6%) allele whereas, in Purulia, AG KAA (24.4%) allele was frequently found after S G KAA (22.2%) and A AKAA (17.8%) in 2008. Wild-SAKAA allele (18.2%) was prevalent in Kolkata followed by AG K AT (17.1%), and A AKAA allele (17.1) in 2009 whereas, in Purulia, AG KAA (30.8%) was most common haplotype after S G KAA (25%) ( Figure 2F ).

Since 2009, no single use of SP has been recommended in India. Instead, in the years that followed, the polymorphism at the dhfr-S108N codon grew and peaked in 2017 in Kolkata and Purulia at 93.6% and 931%, respectively. Similar to this, N51I polymorphism grew over the years and peaked in 2017 at 87.2% in Kolkata and 75.9% in Purulia. Following this, the percentage of pfdhfr-C59R mutation rose to 68.1% in Kolkata and 75.9% in Purulia. Codon A16V showed no mutations ( Figure 2C ).

In addition, pfdhps-S436A and A437G polymorphisms significantly increased from 2010 to 2017 and reached as high as 74.5 and 93.6%, respectively, in Kolkata, whereas 79.3% and 87.9% of isolates, respectively, exhibited mutant-S436A and A437G allele in Purulia. In the following years, the prevalence of the pfdhps-K540E mutation quickly rose, reaching as high as 68.1% in Kolkata and 70.7% in Purulia in 2017. The frequency of the pfdhps-A581G and A613T mutations did not significantly increase ( Figure 2D ).

Since 2010, wild pfdhfr-ANCSI haplotype rapidly reduced in number, whereas most vulnerable triple mutant A IRN I allele was predominantly observed in Kolkata (61.7%) and in Purulia (63.8%) followed by double mutant A I C N I allele (25.5% in Kolkata and 12.1% in Purulia) in 2017. Proportion of single mutant ANC N I allele was gradually decreased in subsequent years. Triple mutant AN RNL allele was observed for first time in 2010 at both study sites and since then it has maintained a low frequency (< 5%) ( Figure 2E ).

Like pfdhfr, frequencies of wild-pfdhps-SAKAA haplotype also reduced drastically since 2010. Triple mutant AGE AA allele was prevalent in Kolkata (31.9%) and in Purulia (37.9%) in 2017 followed by quadruple mutant AGE A T allele (29.8% in Kolkata and 32.8% Purulia). Frequency of single mutant S G KAA and double mutant AG KAA allele were gradually decreased in consequent years. Triple-mutant S GEG A allele was observed initially in 2015 at Kolkata and thereafter maintained a very low frequency (<5%) ( Figure 2F ).

In 2008, the number of SP-treatment failure (ETF) cases was very less in both the study sites, but SP-resistant pattern in Purulia was observed differently from Kolkata. In Purulia, ETFs were associated with double dhfr-A I C N I+ triple dhps- AGE AA mutation (p < 0.05) whereas, in Kolkata, isolates representing triple dhfr-A IRN I +triple dhps- AGE AA mutation were associated with ETF (p < 0.05) in 2008 ( Table 2 ). Both haplotype-combinations showed very high IC₅₀ for pyrimethamine and sulfadoxine and proved as true in vitro resistant isolates ( Figures 4A, B ). In Kolkata, two LTF cases were observed with A I C N I- AG KAA allele. One each LTF case was reported with A I C N I- S G K GT and A IRN I- AGE AA haplotype whereas, in Purulia, one each LTF was found with A I C N I- AGE AA, A I C N I- A AKAT, A I C N I-SAK G A, and ANC N I- AG KAT allele in 2008. Novel dhfr-108I mutation was never observed alone; it was always found with 51I mutation (A I C I I) and possessed very high IC₅₀s for pyrimethamine (p < 0.001). In 2009, 64 out of 65 (98%) ETF in Kolkata and seven of 10 (70%) ETF in Purulia represented novel A I C I I haplotype. ETFs in Kolkata were strongly correlated with A I C I I- AG KA T , A I C I I- AG KAA, and A I C I I-S G K GT combination (pfdhfr-pfdhps) mutations (p < 0.001) whereas, in Purulia, ETFs were frequently found with A I C I I-S G K GT and A I C I I- AG KAA mutations (p < 0.05). Another two ETFs were reported with A I C N I- AG KAA allele in Purulia. On the contrary, in Kolkata, LTFs were frequently found with A I C I I- AG KA T and A IRN I- AGE AA mutations whereas, in Purulia, LTFs were commonly observed with A IRN I- AGE AA, A I C N I- AG KAA, and A I C I I-S G K GT haplotype ( Table 2 ). Above mentioned haplotypes represented higher IC₅₀s for pyrimethamine and sulfadoxine (p < 0.01) ( Figures 4A, B ). In multivariate logistic-regression after adjusting parasitemia, age, and dose of SP, an infection linked with pfdhfr AIC I I-pfdhps S G K GT , pfdhfr AIC I I-pfdhps AG KAA, and pfdhfr AIC I I- pfdhps AG KA T mutation was substantially more likely to have ETF (odds ratio, 94.4 P < 0.0001).

The frequency of wild ANCSI-SAKAA allele subsequently decreased after 2010. The molecular markers associated with in vitro SP-resistance were highly varied in two study sites. In Kolkata, in vitro PYR resistance was highly correlated to pfdhfr-A IRN I and AN RNL allele (p < 0.001), whereas intermediate PYR resistance was associated with A I C N I haplotype (p < 0.001). In Purulia, in vitro PYR resistance was strongly associated with pfdhfr-AN RN I and A IRN I haplotype (p < 0.01), whereas intermediate PYR-resistance was correlated with ANC N I and A I C N I allele (p < 0.001) ( Figure 4C ).

In Kolkata, isolates presenting pfdhps- AGE A T, S G K GT, and AGE AA haplotype possessed very high IC₅₀s for sulfadoxine, proving resistant to sulfadoxine (p < 0.01). Furthermore, in vitro SDX resistance in Purulia was highly correlated with AGE A T and AGE AA haplotype (p < 0.01). In both study sites, single mutant pfdhps-AAKAA, SAKGA, SGKAA, and double mutant pfdhps-AGKAA alleles represented low to moderate IC₅₀s for SDX and were associated with intermediate SDX resistant (p < 0.05) ( Figure 4D ).

Isolates representing A IRN I- AGE AA, A IRN I- AGE A T , and AN RNL - AGE AA combination mutations possessed very high IC₅₀s for PYR and SDX and thereby proving highly resistant to PYR and SDX in vitro (p < 0.001). The isolates containing ANCSI-SAKAA, ANC N I-SAKAA, AN RN I-S G KAA, and A I C N I- AG KAA haplotype represented moderate to high IC₅₀s pyrimethamine and sulfadoxine ( Table 3 ). In multivariate logistic-regression after adjusting parasitemia, age, and dose of SP, an infection linked with pfdhfr A IRN I-pfdhps AGE AA and pfdhfr A IRN I-pfdhps AGE A T mutation was substantially more likely to have an in vitro SP resistance (odds ratio, 89.5, P < 0.001; 93.8, P < 0.001, respectively).