Delta-like 1 ligand (DLL) in cerebrospinal fluid for ...
1
Delta-like 1 ligand (DLL) in cerebrospinal fluid for detection of Mycobacterium tuberculosis meningitis Nathan C Bahr 1,2,3 , Grace Linder 2 , Ryan Halupnick 2 , Reuben Kiggundu 3 , Henry W Nabeta 3 , Darlisha Williams 1,3 , David B Meya 1,3 , Joshua Rhein 1,2,3 , and David R Boulware 1,2 Poster #1703, Session 208 ID Week 2014 Philadelphia, PA Nathan C Bahr MD, MA, ©TropMed [email protected] Abstract Background: Tuberculosis meningitis (TBM) diagnosis is notoriously difficult, new biomarkers are needed to allow for improved diagnostic accuracy. We evaluated the diagnostic utility of a novel biomarker, delta-like ligand 1 (DLL1), a Notch ligand, which selectively drives antigen-specific CD4 T helper1 cell responses. DLL1 polymorphisms increase susceptibility to other intraceullar (Th1) organisms (e.g. Leishmaniasis). Methods: CSF DLL1 concentrations were measured by ELISA in 136 patients with suspected meningitis, of which 21 patients had TBM, 71 patients cryptococcal meningitis (CM), and 44 patients tested negative for both CM and TBM (termed ‘other’). TBM was diagnosed either by GeneXpert MTB/Rif assay (Cepheid, Sunnyvale, CA) and/or Bactec MGIT culture on CSF. We evaluated the diagnostic performance of DLL1 for TBM. Results: Patient characteristics were similar at diagnosis except for serum CD4 (lower in CM than other groups), percent male (lower in ‘other’ meningitis than other groups) and CSF WBC (greater in TBBM than other groups). Mean DLL1 CSF concentrations were significantly higher in patients with TBM (1197 pg/mL; 95%CI: 603-1791 pg/mL) than CM (450 pg/mL; 95%CI: 404-497 pg/mL) or ‘other’ aseptic meningitis (511 pg/mL; 95%CI: 359-662 pg/mL). A cutoff of >600pg/mL in CSF for TBM had 71% sensitivity, 73% specificity, 33% positive predictive value (PPV), and 93% negative predictive value (NPV) 94% (AUC=0.794). As the DLL1 level increased, the likelihood of TBM increased with specificity of 91% and PPV of 50% above >800 pg/mL. Conclusions: CSF DLL1 exhibited good diagnostic performance, and may have a role as a low cost adjunctive diagnostic tool for TBM. Misclassification bias (of non-detection of TBM classified as ‘other’) hampers diagnostic studies, and larger studies are required in the future. Background •Individuals with TB and HIV co-infection have a higher rate of TBM and a higher case mortality than persons without HIV infection. •In settings where TB-HIV co-infection is common, mortality is often > 50% and those who do survive may have significant neurologic disability. Mortality due to TBM is high in large part due to difficulty in diagnosis. •Ziehl-Neelsen staining for acid-fast bacilli (AFB) is rapid but sensitivity is poor (usually 10-20%) while culture is more sensitive but slow growth (2-6 weeks depending on the media used) precludes rapid clinical decision-making and sensitivity is still only 60-70% sensitive. •A rapid, accurate diagnostic method of detecting Mycobacterium tuberculosis in the cerebrospinal fluid (CSF) would likely lead to improved patient outcomes. •Delta-like 1 protein (DLL1) is a trans-membrane protein and part of the Notch-ligand family and plays a role in fatty acid metabolism which is important for the Mycobacterium tuberculosis cell wall and; •DLL1 has been found to selectively drive antigen-specific CD4 T helper1 cell responses and DLL1 polymorphisms increase susceptibility to other intracellular (Th1) organisms such as and may drive a similar Th1 response in patients infected with Mycobacterium tuberculosis. Results Conclusions/Implications: • CSF DLL1 levels are significantly higher in patients with tuberculosis meningitis than in patients with either cryptococcal meningitis or aseptic meningitis (either considered separately or combined) • DLL1 shows moderate accuracy in identifying microbiologically proven tuberculosis meningitis • DLL1 may have a role as a rapid (< 1 day) test in its current ELISA form for diagnosis of TBM • Further, larger studies, potentially combining this test with other tests should be done to further explore the potential role of DLL1 in diagnosis of tuberculosis meningitis diagnostics Methods • Patients or surrogates provided informed consent and IRB approvals were obtained. • Cryptococcal meningitis was diagnosed by cryptococcal antigen (CRAG) lateral flow assay (IMMY Inc, Norman, OK, USA). • Tuberculosis microbiologic diagnosis was obtained by polymerase chain reaction via the GeneXpert system using the MTB/Rif cartridge (Cepheid Inc, Sunnyvale, CA, USA). • CSF DLL1 concentrations were measured using the human soluble DLL1 ELISA kit (AdipoGen Inc., Incheon, Korea) according to manufacturer instructions using cryopreserved specimens (-80°C). • Testing kits were purchased by the investigators, the manufacturers had no role in this study. • Statistical analysis was done using SPSS version 21 (IBM Inc., Armonk, NY, USA). • Clinical and demographic data was grouped by diagnosis (TBM, cryptococcal, or other) and compared using ANOVA, independent samples t tests, and chi-square tests where appropriate. • ANOVA and independent samples t tests were used to compare mean DLL1 concentrations and area under the curve was calculated using a receiver operator characteristic curve. Study Population: •HIV infected individuals suspected to have meningitis in Kampala, Uganda at Mulago National Referral Hospital. •Enrolled as part of either the Cryptococcal Optimal ART Timing (COAT) trial (clinicaltrials.gov identifier NCT01075152) or the Adjunctive Sertraline for the Treatment of HIV-Associated Cryptococcal Meningitis (ASTRO-CM) trial (clinicaltrials.gov identifier NCT01802385) . Acknowledgements: This research was supported via the Fogarty International Center (R25TW009345), the National Institute of Neurologic Diseases and Stroke (R01NS086312) and the National Institute of Allergy and Infectious Diseases (U01AI089244). The authors also thank Dr. Abdu Musubire for clinical care. Cryptococcal Meningitis (n=71) Tuberculosis Meningitis (n=21) Other / Aseptic Meningitis (n=44) P value Sex, %Male 69% 57% 41% 0.012 Age in years 37 (25-39) 38 (35-41) 36 (33-40) 0.827 CSF Protein, mg/ dL a,b, 99 (76-122) 152 (105-199) 103 (67-139) 0.106 CSF White Blood Cells, cells/μL a,b 70 (32-108) 192 (59-325) 52 (8-95) 0.011 CSF % Lymphocytes a,b,c 90% (79-100%) 94% (80-100%) 75% (55-95%) 0.174 % Alive at Hospital Discharge b,c 72% 65% 84% 0.247 CD4 T Cell Count cells/μL a,b,c 41 (27-54) 95 (41-148) 85 (31-140) 0.010 Headache Duration, days a,b,c 20 (16-25) 21 (11-32) 19 (13-26) 0.926 CSF DLL1, pg/mL 450 (404-497) 1197 (603-1791) 511 (359-662) <0.001 Table 1: Patient Characteristics Patient characteristics compared between patients with cryptococcal meningitis, tuberculosis meningitis, and patients clinically suspected to have meningitis but found to have no diagnostic confirmation of either cryptococcal meningitis or tuberculosis meningitis (other/aseptic meningitis). Numeric values given in means with 95% confidence intervals. a: For cryptococcal meningitis: n=67 for CSF protein, n=66 for WBC, n=25 for % lymphocytes when CSF WBC > 5 cells/ μL, n=68 for CD4, n=70 for headache duration. b: For TB meningitis: n=16 for % lymphocytes when CSF WBC > 5 cells/μL, n=17 for % alive at discharge, n=12 for CD4, n=19 for headache duration. c: For other/aseptic meningitis: n=40 for CSF protein, n=42 for WBC, n=14 for % lymphocytes when CSF WBC > 5 cells/ μL, n=37 for % alive at discharge, n=17 for CD4, 40 for headache duration. DLL1 Cutoff (pg/mL) Sensitivity Specificity Positive Predictive Value Negative Predictive Value >500 76% (16/21) 66% (76/115) 29% (16/55) 94% (76//81) >600 71% (15/21) 73% (84/115) 33% (15/46) 93% (84/90) >700 52% (11/21) 85% (98/115) 39% (11/28) 91% (98/108) >800 48% (10/21) 91% (105/115) 50% (10/20) 91% (105/116) Table 2: Test performance for detection of TB meningitis by DLL1 value 1 Division of Infectious Disease and International Medicine, Department of Medicine, University of Minnesota, Minneapolis, USA; 2 Center for Infectious Disease & Microbiology Translational Research, Minneapolis, USA. 3 Infectious Disease Institute, Makerere University, Kampala, Uganda AUC: 0.778 Figure 1: ROC curve to evaluate DLL1 performance for detection of TBM on CSF
Transcript of Delta-like 1 ligand (DLL) in cerebrospinal fluid for ...
Delta-like 1 ligand (DLL) in cerebrospinal fluid for detection of Mycobacterium tuberculosis meningitis
Nathan C Bahr1,2,3, Grace Linder2, Ryan Halupnick2, Reuben Kiggundu3, Henry W Nabeta3, Darlisha Williams1,3, David B Meya1,3, Joshua Rhein1,2,3, and David R Boulware1,2
Poster #1703, Session 208 ID Week 2014
Philadelphia, PA
Nathan C Bahr MD, MA, ©TropMed [email protected]
Abstract Background: Tuberculosis meningitis (TBM) diagnosis is notoriously difficult, new biomarkers are needed to allow for improved diagnostic accuracy. We evaluated the diagnostic utility of a novel biomarker, delta-like ligand 1 (DLL1), a Notch ligand, which selectively drives antigen-specific CD4 T helper1 cell responses. DLL1 polymorphisms increase susceptibility to other intraceullar (Th1) organisms (e.g. Leishmaniasis). Methods: CSF DLL1 concentrations were measured by ELISA in 136 patients with suspected meningitis, of which 21 patients had TBM, 71 patients cryptococcal meningitis (CM), and 44 patients tested negative for both CM and TBM (termed ‘other’). TBM was diagnosed either by GeneXpert MTB/Rif assay (Cepheid, Sunnyvale, CA) and/or Bactec MGIT culture on CSF. We evaluated the diagnostic performance of DLL1 for TBM. Results: Patient characteristics were similar at diagnosis except for serum CD4 (lower in CM than other groups), percent male (lower in ‘other’ meningitis than other groups) and CSF WBC (greater in TBBM than other groups). Mean DLL1 CSF concentrations were significantly higher in patients with TBM (1197 pg/mL; 95%CI: 603-1791 pg/mL) than CM (450 pg/mL; 95%CI: 404-497 pg/mL) or ‘other’ aseptic meningitis (511 pg/mL; 95%CI: 359-662 pg/mL). A cutoff of >600pg/mL in CSF for TBM had 71% sensitivity, 73% specificity, 33% positive predictive value (PPV), and 93% negative predictive value (NPV) 94% (AUC=0.794). As the DLL1 level increased, the likelihood of TBM increased with specificity of 91% and PPV of 50% above >800 pg/mL. Conclusions: CSF DLL1 exhibited good diagnostic performance, and may have a role as a low cost adjunctive diagnostic tool for TBM. Misclassification bias (of non-detection of TBM classified as ‘other’) hampers diagnostic studies, and larger studies are required in the future.
Background
• Individuals with TB and HIV co-infection have a higher rate of TBM and a higher case mortality than persons without HIV infection.
• In settings where TB-HIV co-infection is common, mortality is often > 50% and those who do survive may have significant neurologic disability. Mortality due to TBM is high in large part due to difficulty in diagnosis.
• Ziehl-Neelsen staining for acid-fast bacilli (AFB) is rapid but sensitivity is poor (usually 10-20%) while culture is more sensitive but slow growth (2-6 weeks depending on the media used) precludes rapid clinical decision-making and sensitivity is still only 60-70% sensitive.
• A rapid, accurate diagnostic method of detecting Mycobacterium tuberculosis in the cerebrospinal fluid (CSF) would likely lead to improved patient outcomes.
• Delta-like 1 protein (DLL1) is a trans-membrane protein and part of the Notch-ligand family and plays a role in fatty acid metabolism which is important for the Mycobacterium tuberculosis cell wall and;
• DLL1 has been found to selectively drive antigen-specific CD4 T helper1 cell responses and DLL1 polymorphisms increase susceptibility to other intracellular (Th1) organisms such as and may drive a similar Th1 response in patients infected with Mycobacterium tuberculosis.
Results
Conclusions/Implications: • CSF DLL1 levels are significantly higher in patients with tuberculosis meningitis than in patients with
either cryptococcal meningitis or aseptic meningitis (either considered separately or combined) • DLL1 shows moderate accuracy in identifying microbiologically proven tuberculosis meningitis • DLL1 may have a role as a rapid (<1 day) test in its current ELISA form for diagnosis of TBM • Further, larger studies, potentially combining this test with other tests should be done to further
explore the potential role of DLL1 in diagnosis of tuberculosis meningitis diagnostics
Methods • Patients or surrogates provided informed consent and IRB approvals were obtained.
• Cryptococcal meningitis was diagnosed by cryptococcal antigen (CRAG) lateral flow assay (IMMY Inc, Norman, OK, USA).
• Tuberculosis microbiologic diagnosis was obtained by polymerase chain reaction via the GeneXpert system using the MTB/Rif cartridge (Cepheid Inc, Sunnyvale, CA, USA).
• CSF DLL1 concentrations were measured using the human soluble DLL1 ELISA kit (AdipoGen Inc., Incheon, Korea) according to manufacturer instructions using cryopreserved specimens (-80°C).
• Testing kits were purchased by the investigators, the manufacturers had no role in this study.
• Statistical analysis was done using SPSS version 21 (IBM Inc., Armonk, NY, USA).
• Clinical and demographic data was grouped by diagnosis (TBM, cryptococcal, or other) and compared using ANOVA, independent samples t tests, and chi-square tests where appropriate.
• ANOVA and independent samples t tests were used to compare mean DLL1 concentrations and area under the curve was calculated using a receiver operator characteristic curve.
Study Population: • HIV infected individuals suspected to have meningitis in Kampala, Uganda at Mulago National Referral Hospital. • Enrolled as part of either the Cryptococcal Optimal ART Timing (COAT) trial (clinicaltrials.gov identifier NCT01075152) or the Adjunctive Sertraline for the Treatment of HIV-Associated Cryptococcal Meningitis (ASTRO-CM) trial (clinicaltrials.gov identifier NCT01802385) .
Acknowledgements: This research was supported via the Fogarty International Center (R25TW009345), the National Institute of
Neurologic Diseases and Stroke (R01NS086312) and the National Institute of Allergy and Infectious Diseases (U01AI089244). The authors also thank Dr. Abdu Musubire for clinical care.
Cryptococcal
Meningitis (n=71)
Tuberculosis
Meningitis (n=21)
Other / Aseptic
Meningitis (n=44) P value
Sex, %Male 69% 57% 41% 0.012
Age in years 37 (25-39) 38 (35-41) 36 (33-40) 0.827
CSF Protein, mg/dL a,b, 99 (76-122) 152 (105-199) 103 (67-139) 0.106
CSF White Blood Cells, cells/µL a,b 70 (32-108) 192 (59-325) 52 (8-95) 0.011
CSF % Lymphocytes a,b,c 90% (79-100%) 94% (80-100%) 75% (55-95%) 0.174
% Alive at Hospital Discharge b,c 72% 65% 84% 0.247
CD4 T Cell Count cells/µL a,b,c 41 (27-54) 95 (41-148) 85 (31-140) 0.010
Headache Duration, days a,b,c 20 (16-25) 21 (11-32) 19 (13-26) 0.926
CSF DLL1, pg/mL 450 (404-497) 1197 (603-1791) 511 (359-662) <0.001
Table 1: Patient Characteristics
Patient characteristics compared between patients with cryptococcal meningitis, tuberculosis meningitis, and patients clinically suspected to have meningitis but found to have no diagnostic confirmation of either cryptococcal meningitis or tuberculosis meningitis (other/aseptic meningitis). Numeric values given in means with 95% confidence intervals.
a: For cryptococcal meningitis: n=67 for CSF protein, n=66 for WBC, n=25 for % lymphocytes when CSF WBC >5 cells/µL, n=68 for CD4, n=70 for headache duration. b: For TB meningitis: n=16 for % lymphocytes when CSF WBC >5 cells/µL, n=17 for % alive at discharge, n=12 for CD4, n=19 for headache duration. c: For other/aseptic meningitis: n=40 for CSF protein, n=42 for WBC, n=14 for % lymphocytes when CSF WBC >5 cells/µL, n=37 for % alive at discharge, n=17 for CD4, 40 for headache duration.
DLL1 Cutoff (pg/mL) Sensitivity Specificity
Positive Predictive
Value
Negative Predictive
Value
>500 76% (16/21) 66% (76/115) 29% (16/55) 94% (76//81)
>600 71% (15/21) 73% (84/115) 33% (15/46) 93% (84/90)
>700 52% (11/21) 85% (98/115) 39% (11/28) 91% (98/108)
>800 48% (10/21) 91% (105/115) 50% (10/20) 91% (105/116)
Table 2: Test performance for detection of TB meningitis by DLL1 value
1 Division of Infectious Disease and International Medicine, Department of Medicine, University of Minnesota, Minneapolis, USA; 2 Center for Infectious Disease & Microbiology Translational Research, Minneapolis, USA. 3 Infectious Disease Institute, Makerere University, Kampala, Uganda
AUC: 0.778
Figure 1: ROC curve to evaluate DLL1 performance for detection of TBM on CSF
