Journal of the American Medical Informatics Association, Feb 9, 2022
ObjectiveThis study aims to establish an informative dynamic prediction model of treatment outcom... more ObjectiveThis study aims to establish an informative dynamic prediction model of treatment outcomes using follow-up records of tuberculosis (TB) patients, which can timely detect cases when the current treatment plan may not be effective.Materials and MethodsWe used 122 267 follow-up records from 17 958 new cases of pulmonary TB in the Republic of Moldova. A dynamic prediction framework integrating landmark modeling and machine learning algorithms was designed to predict patient outcomes during the course of treatment. Sensitivity and positive predictive value (PPV) were calculated to evaluate performance of the model at critical time points. New measures were defined to determine when follow-up laboratory tests should be conducted to obtain most informative results.ResultsThe random-forest algorithm performed better than support vector machine and penalized multinomial logistic regression models for predicting TB treatment outcomes. For all 3 outcome classes (ie, cured, not cured, and died after 24 months following treatment initiation), sensitivity and PPV of prediction models improved as more follow-up information was collected. Specifically, sensitivity and PPV increased from 0.55 to 0.84 and from 0.32 to 0.88, respectively, for the not cured class.ConclusionThe dynamic prediction framework utilizes longitudinal laboratory test results to predict patient outcomes at various landmarks. Sputum culture and smear results are among the important variables for prediction; however, the most recent sputum result is not always the most informative one. This framework can potentially facilitate a more effective treatment monitoring program and provide insights for policymakers toward improved guidelines on follow-up tests.
Transmission-driven multi-/extensively drug resistant (M/XDR) tuberculosis (TB) is the largest si... more Transmission-driven multi-/extensively drug resistant (M/XDR) tuberculosis (TB) is the largest single contributor to human mortality due to antimicrobial resistance. A few major clades of the Mycobacterium tuberculosis complex belonging to lineage 2, responsible for high prevalence of MDR-TB in Eurasia, show outstanding transnational distributions. Here, we determined factors underlying the emergence and epidemic spread of the W148 clade by genome sequencing and Bayesian demogenetic analyses of 720 isolates from 23 countries. We dated a common ancestor around 1963 and identified two successive epidemic expansions in the late 1980s and late 1990s, coinciding with major socio-economic changes in the post-Soviet Era. These population expansions favored accumulation of resistance mutations to up to 11 anti-TB drugs, with MDR evolving toward additional resistances to fluoroquinolones and second-line injectable drugs within 20 years on average. Timescaled haplotypic density analysis revea...
International Journal of Tuberculosis and Lung Disease, Dec 1, 2015
Nosocomial transmission of multidrug-resistant tuberculosis (MDR-TB) was ascertained by 24-locus ... more Nosocomial transmission of multidrug-resistant tuberculosis (MDR-TB) was ascertained by 24-locus mycobacterial interspersed repetitive units-variable number of tandem repeats (MIRU-VNTR) and spoligotyping at four hospitals in the Republic of Moldova, a high MDR-TB burden country. Overall, 5.1% of patients with pan-susceptible TB at baseline were identified with MDR-TB during in-patient treatment. In 75% of cases, the MDR-TB strain was genetically distinct from the non-MDR-TB strain at baseline, suggesting a high rate of nosocomial transmission of MDR-TB. The highest proportion (40.3%) of follow-up MDR-TB isolates was associated with the M. tuberculosis URAL 163-15 strain.
BackgroundAs part of its End TB strategy, the WHO has identified the need for non-sputum-based di... more BackgroundAs part of its End TB strategy, the WHO has identified the need for non-sputum-based diagnostics that meet target product profiles (TPP) of 90% sensitivity and 70% specificity for diagnosis of Active TB (ATB) and 75% sensitivity and specificity for predicting progression from Latent TB (LTB) to ATB. The successful translation of a 3-gene blood-based signature, identified using diverse datasets, into a prototype point-of-care diagnostic, that meets the WHO TPPs, has demonstrated the power of integrating large amounts of heterogeneous data to identify generalizable disease signatures. We hypothesized that integration of more diverse datasets, comprising patients with ATB or other inflammatory lung diseases, would identify novel, robust signatures, for diagnosing ATB and predicting progression from LTB to ATB, that meet the WHO TPPs.MethodsUsing multi-cohort analyses, we integrated and analyzed data from 3,615 peripheral blood samples, in 49 publicly available transcriptomic datasets (discovery cohorts), from healthy controls and patients with LTB, ATB, and other diseases (COPD, viral infections, sarcoidosis, etc.). We used data from (1) 3,836 blood samples in 28 retrospective datasets and (2) 360 prospectively collected blood samples, from a household contact study in Moldova, as validation cohorts.ResultsUsing the discovery cohorts we identified a 9-gene signature for diagnosing ATB patients from healthy controls, or individuals with LTB or other diseases. The signature achieved 90% sensitivity and 82% specificity in retrospective validation (Figure 1A) and 90% sensitivity and 69% specificity in the prospective cohort from Moldova (Figure 1B). In a longitudinal cohort of adolescents, the 9-gene signature predicted progression from LTB to ATB up to 1 year prior to sputum conversion with 76% sensitivity and 83% specificity (Figure 1C). Finally, the signature predicted prolonged lung inflammation post-treatment in the Catalysis Treatment Response Cohort (Figure 1D).9-gene TB signature validates in independent retrospective and prospective cohorts.Open in a separate window(A) ROC curves for comparing ATB vs. all other samples (All), or individually comparing ATB vs. Healthy, LTBI or Other Disease (OD) samples in independent retrospective validation and (B) a prospective Moldova cohort. (C) ROC curves comparing progressor and non-progressor samples collected at different time points in the Adolescent Cohort Study (ACS), for the 9-gene signature in solid lines and a previous 3-gene signature in dashed lines, (D) Distribution of the 9-gene score at different time points post-treatment in the Catalysis Treatment Response Cohort Study, for individuals with persistent lung inflammation at 24 weeks compared with those who had clear lungs at 24 weeks.ConclusionOverall, the 9-gene signature meets the WHO TPPs required for the End TB strategy.Disclosures Purvesh Khatri, PhD, Cepheid, Inc.: Advisor/Consultant|Inflammatix, Inc.: Advisor/Consultant|Inflammatix, Inc.: Inflammatix is in negotiations to license the 9-gene signature discussed here.|Inflammatix, Inc.: Stocks/Bonds.
Tuberculosis caused by Mycobacterium tuberculosis is one of the leading causes of death from a si... more Tuberculosis caused by Mycobacterium tuberculosis is one of the leading causes of death from a single infectious agent. Identifying dominant epitopes and comparing their reactivity in different tuberculosis (TB) infection states can help design diagnostics and vaccines. We performed a proteome-wide screen of 20,610 Mtb derived peptides in 21 Active TB (ATB) patients 3-4 months post-diagnosis of pulmonary TB (mid-treatment) using an IFNgamma and IL-17 Fluorospot assay. Responses were mediated exclusively by IFNgamma and identified a total of 137 unique epitopes, with each patient recognizing, on average, 8 individual epitopes and 22 epitopes (16%) recognized by 2 or more participants. Responses were predominantly directed against antigens part of the cell wall and cell processes category. Testing 517 peptides spanning TB vaccine candidates and ESAT-6 and CFP10 antigens also revealed differential recognition between ATB participants mid-treatment and healthy IGRA+ participants of seve...
Bulletin of the Academy of Sciences of Moldova. Medical Sciences
Tuberculosis, one of the ancient infections that especially affects humans, the main target organ... more Tuberculosis, one of the ancient infections that especially affects humans, the main target organ of this disease being the lung, is caused by M. tuberculosis and remains one of the most important public health problems, especially in developing countries. The emergence of multidrug-resistant and extensively drug-resistant tuberculosis is considered a serious threat to tuberculosis control worldwide. The burden of drug-resistant tuberculosis remains increasing in countries endemic for this infection. The risk factors that influence this situation and that are elucidated more frequently in the specialized literature are the abusive use of drugs, the failure of the therapeutic regimen, the relapse of tuberculosis, the difficulty of obtaining drug sensitivity testing in a fast time. In Moldova, but also in other regions of Eastern Europe, drug resistance represents a serious obstacle in the effective control of this disease. There is evidence that the transmission of resistant strains,...
While the goal of universal drug susceptibility testing has been a key component of the WHO End T... more While the goal of universal drug susceptibility testing has been a key component of the WHO End TB Strategy, in practice, this remains inaccessible to many. Rapid molecular tests for tuberculosis (TB) and antituberculosis drug resistance could significantly improve access to testing.
Journal of the American Medical Informatics Association, Feb 9, 2022
ObjectiveThis study aims to establish an informative dynamic prediction model of treatment outcom... more ObjectiveThis study aims to establish an informative dynamic prediction model of treatment outcomes using follow-up records of tuberculosis (TB) patients, which can timely detect cases when the current treatment plan may not be effective.Materials and MethodsWe used 122 267 follow-up records from 17 958 new cases of pulmonary TB in the Republic of Moldova. A dynamic prediction framework integrating landmark modeling and machine learning algorithms was designed to predict patient outcomes during the course of treatment. Sensitivity and positive predictive value (PPV) were calculated to evaluate performance of the model at critical time points. New measures were defined to determine when follow-up laboratory tests should be conducted to obtain most informative results.ResultsThe random-forest algorithm performed better than support vector machine and penalized multinomial logistic regression models for predicting TB treatment outcomes. For all 3 outcome classes (ie, cured, not cured, and died after 24 months following treatment initiation), sensitivity and PPV of prediction models improved as more follow-up information was collected. Specifically, sensitivity and PPV increased from 0.55 to 0.84 and from 0.32 to 0.88, respectively, for the not cured class.ConclusionThe dynamic prediction framework utilizes longitudinal laboratory test results to predict patient outcomes at various landmarks. Sputum culture and smear results are among the important variables for prediction; however, the most recent sputum result is not always the most informative one. This framework can potentially facilitate a more effective treatment monitoring program and provide insights for policymakers toward improved guidelines on follow-up tests.
Transmission-driven multi-/extensively drug resistant (M/XDR) tuberculosis (TB) is the largest si... more Transmission-driven multi-/extensively drug resistant (M/XDR) tuberculosis (TB) is the largest single contributor to human mortality due to antimicrobial resistance. A few major clades of the Mycobacterium tuberculosis complex belonging to lineage 2, responsible for high prevalence of MDR-TB in Eurasia, show outstanding transnational distributions. Here, we determined factors underlying the emergence and epidemic spread of the W148 clade by genome sequencing and Bayesian demogenetic analyses of 720 isolates from 23 countries. We dated a common ancestor around 1963 and identified two successive epidemic expansions in the late 1980s and late 1990s, coinciding with major socio-economic changes in the post-Soviet Era. These population expansions favored accumulation of resistance mutations to up to 11 anti-TB drugs, with MDR evolving toward additional resistances to fluoroquinolones and second-line injectable drugs within 20 years on average. Timescaled haplotypic density analysis revea...
International Journal of Tuberculosis and Lung Disease, Dec 1, 2015
Nosocomial transmission of multidrug-resistant tuberculosis (MDR-TB) was ascertained by 24-locus ... more Nosocomial transmission of multidrug-resistant tuberculosis (MDR-TB) was ascertained by 24-locus mycobacterial interspersed repetitive units-variable number of tandem repeats (MIRU-VNTR) and spoligotyping at four hospitals in the Republic of Moldova, a high MDR-TB burden country. Overall, 5.1% of patients with pan-susceptible TB at baseline were identified with MDR-TB during in-patient treatment. In 75% of cases, the MDR-TB strain was genetically distinct from the non-MDR-TB strain at baseline, suggesting a high rate of nosocomial transmission of MDR-TB. The highest proportion (40.3%) of follow-up MDR-TB isolates was associated with the M. tuberculosis URAL 163-15 strain.
BackgroundAs part of its End TB strategy, the WHO has identified the need for non-sputum-based di... more BackgroundAs part of its End TB strategy, the WHO has identified the need for non-sputum-based diagnostics that meet target product profiles (TPP) of 90% sensitivity and 70% specificity for diagnosis of Active TB (ATB) and 75% sensitivity and specificity for predicting progression from Latent TB (LTB) to ATB. The successful translation of a 3-gene blood-based signature, identified using diverse datasets, into a prototype point-of-care diagnostic, that meets the WHO TPPs, has demonstrated the power of integrating large amounts of heterogeneous data to identify generalizable disease signatures. We hypothesized that integration of more diverse datasets, comprising patients with ATB or other inflammatory lung diseases, would identify novel, robust signatures, for diagnosing ATB and predicting progression from LTB to ATB, that meet the WHO TPPs.MethodsUsing multi-cohort analyses, we integrated and analyzed data from 3,615 peripheral blood samples, in 49 publicly available transcriptomic datasets (discovery cohorts), from healthy controls and patients with LTB, ATB, and other diseases (COPD, viral infections, sarcoidosis, etc.). We used data from (1) 3,836 blood samples in 28 retrospective datasets and (2) 360 prospectively collected blood samples, from a household contact study in Moldova, as validation cohorts.ResultsUsing the discovery cohorts we identified a 9-gene signature for diagnosing ATB patients from healthy controls, or individuals with LTB or other diseases. The signature achieved 90% sensitivity and 82% specificity in retrospective validation (Figure 1A) and 90% sensitivity and 69% specificity in the prospective cohort from Moldova (Figure 1B). In a longitudinal cohort of adolescents, the 9-gene signature predicted progression from LTB to ATB up to 1 year prior to sputum conversion with 76% sensitivity and 83% specificity (Figure 1C). Finally, the signature predicted prolonged lung inflammation post-treatment in the Catalysis Treatment Response Cohort (Figure 1D).9-gene TB signature validates in independent retrospective and prospective cohorts.Open in a separate window(A) ROC curves for comparing ATB vs. all other samples (All), or individually comparing ATB vs. Healthy, LTBI or Other Disease (OD) samples in independent retrospective validation and (B) a prospective Moldova cohort. (C) ROC curves comparing progressor and non-progressor samples collected at different time points in the Adolescent Cohort Study (ACS), for the 9-gene signature in solid lines and a previous 3-gene signature in dashed lines, (D) Distribution of the 9-gene score at different time points post-treatment in the Catalysis Treatment Response Cohort Study, for individuals with persistent lung inflammation at 24 weeks compared with those who had clear lungs at 24 weeks.ConclusionOverall, the 9-gene signature meets the WHO TPPs required for the End TB strategy.Disclosures Purvesh Khatri, PhD, Cepheid, Inc.: Advisor/Consultant|Inflammatix, Inc.: Advisor/Consultant|Inflammatix, Inc.: Inflammatix is in negotiations to license the 9-gene signature discussed here.|Inflammatix, Inc.: Stocks/Bonds.
Tuberculosis caused by Mycobacterium tuberculosis is one of the leading causes of death from a si... more Tuberculosis caused by Mycobacterium tuberculosis is one of the leading causes of death from a single infectious agent. Identifying dominant epitopes and comparing their reactivity in different tuberculosis (TB) infection states can help design diagnostics and vaccines. We performed a proteome-wide screen of 20,610 Mtb derived peptides in 21 Active TB (ATB) patients 3-4 months post-diagnosis of pulmonary TB (mid-treatment) using an IFNgamma and IL-17 Fluorospot assay. Responses were mediated exclusively by IFNgamma and identified a total of 137 unique epitopes, with each patient recognizing, on average, 8 individual epitopes and 22 epitopes (16%) recognized by 2 or more participants. Responses were predominantly directed against antigens part of the cell wall and cell processes category. Testing 517 peptides spanning TB vaccine candidates and ESAT-6 and CFP10 antigens also revealed differential recognition between ATB participants mid-treatment and healthy IGRA+ participants of seve...
Bulletin of the Academy of Sciences of Moldova. Medical Sciences
Tuberculosis, one of the ancient infections that especially affects humans, the main target organ... more Tuberculosis, one of the ancient infections that especially affects humans, the main target organ of this disease being the lung, is caused by M. tuberculosis and remains one of the most important public health problems, especially in developing countries. The emergence of multidrug-resistant and extensively drug-resistant tuberculosis is considered a serious threat to tuberculosis control worldwide. The burden of drug-resistant tuberculosis remains increasing in countries endemic for this infection. The risk factors that influence this situation and that are elucidated more frequently in the specialized literature are the abusive use of drugs, the failure of the therapeutic regimen, the relapse of tuberculosis, the difficulty of obtaining drug sensitivity testing in a fast time. In Moldova, but also in other regions of Eastern Europe, drug resistance represents a serious obstacle in the effective control of this disease. There is evidence that the transmission of resistant strains,...
While the goal of universal drug susceptibility testing has been a key component of the WHO End T... more While the goal of universal drug susceptibility testing has been a key component of the WHO End TB Strategy, in practice, this remains inaccessible to many. Rapid molecular tests for tuberculosis (TB) and antituberculosis drug resistance could significantly improve access to testing.
Uploads
Papers by Valeriu Crudu