There are two starting points for this algorithm:
• For a patient who has RIF-susceptible TB by DST, detected by either a molecular (e.g. Xpert MTB/RIF, Xpert Ultra or Truenat) or phenotypic test, but no results are available for INH and the patient is at a high risk for Hr-TB, start at Step 1.
• For a patient who has had an initial TB test that included RIF and INH results (e.g. a moderate complexity automated NAAT was used) in Algorithm 1, skip to Step 4.
1. Collect a good-quality specimen and transport it to the testing laboratory for molecular or phenotypic testing for INH resistance.
a. Testing could follow a two-step process: detection of INH resistance followed by detection of FQ resistance. The two-step process is applicable when a moderate complexity automated NAAT or FL-LPA is used for Hr-TB detection and followed by the low complexity automated NAAT or SL-LPA for FQ resistance detection. A single step option is now available using the first in class low complexity automated NAAT which detects both INH and FQ resistance simultaneously.
b. Phenotypic testing may be required for INH resistance determination as the sensitivity, depending on the test used, may miss ~15% of resistant samples (Table 3.3 in Section 3). Phenotypic testing will be relevant when the patient is at high risk for Hr-TB. If both molecular and phenotypic tests are performed, initiate the tests in parallel; do not wait for the results of one test before initiating the other test.
c. Culture-based phenotypic DST for INH requires 3-8 weeks to produce a result. Phenotypic DST may be useful for evaluating patients with a susceptible mWRD INH result, particularly in populations with a high pretest probability for resistance to INH.
2. If INH resistance is not detected 4A, continue treatment with a first-line regimen in accordance with national guidelines:
a. Conduct additional DST in accordance with national guidelines.
b. Additional molecular or phenotypic DST for resistance to INH may be requested if the patient is considered to be at risk of having Hr-TB, despite the mWRD result.
3. If INH resistance is detected 4B:
a. Using the low complexity automated NAAT will provide simultaneous detection of resistance to INH and FQ. If FQ resistance is not detected follow Steps 3b and 5. If FQ resistance is detected follow Steps 3c iii and 5. If the FQ result is unknown or unsuccessful follow Step 3c.
b. Initiate treatment with an Hr-TB regimen (63):
i. There is no clear evidence showing that adding INH at the usual doses adds benefits or harms to patients. For patient convenience and ease of administration, the four-drug INH/RIF/EMB/PZA (HREZ) fixed-dose combination tablets may be used to deliver the Hr-TB regimen alongside LFX.
ii. According to emerging evidence, patients infected with strains with only inhA promoter mutations, and corresponding modest minimal inhibitory concentration (MIC) increases, may benefit from high-dose INH therapy. Thus, additional INH up to a maximum dose of 15 mg/kg per day may be considered for use with the Hr-TB regimen for such isolates. The added value of isoniazid in the regimen, even when used at the higher dose, declines as MICs increase further.
c. Refer a specimen from a patient with laboratory-confirmed Hr-TB for molecular (e.g. low complexity automated NAAT or SL-LPA) or phenotypic DST for FQ and PZA. (Note that if the Xpert MTB/XDR test was used in Step 1, the FQ result will already be available, so skip to Step 3d.)
i. Rapid molecular testing for FQ resistance is preferred. When used for direct testing of sputum specimens from patients, the low complexity automated NAAT and SL-LPA detects 93% and 86% of patients with FQ resistance, respectively (Table 3.4 in Section 3).
1. Low complexity automated NAATs provide rapid results and are suitable for use at the peripheral level. The first in class test, Xpert MTB/XDR, reports low-level FQ resistance when the mutations gyrA A90V, gyrA S91P and gyrA D94A are detected from the probe melting temperature (64). Phenotypic DST at the clinical breakpoint for moxifloxacin should be performed to confirm the potential value of high-dose moxifloxacin treatment for such patients.
2. The diagnostic accuracy of SL-LPA is similar when it is performed directly on sputum or from cultured isolates. SL-LPA can be used with smear-positive or smear-negative specimens, although a higher indeterminate rate will occur when testing smear-negative specimens.
3. Despite good specificity and sensitivity of low complexity automated NAATs and SL-LPA for the detection of FQ resistance, culture and phenotypic DST are required to exclude resistance to individual FQs completely. In particular, phenotypic DST may be needed in settings with a high pretest probability for resistance to FQ, to exclude resistance when the SL-LPA does not detect mutations associated with resistance.
d. Review FQ resistance results.
i. If FQ resistance is not detected, continue treatment with the LFX-containing Hr-TB regimen.
ii. If FQ resistance is detected:
1. Discontinue use of LFX and change to a 6-month regimen of (INH)/RIF/EMB/PZA (i.e. 6(H) REZ, where the "(H)" indicates that the INH is optional) or an individualized Hr-TB regimen.
2. Refer a specimen for PZA DST if reliable PZA DST is available in the country. Options include the high complexity reverse hybridization NAAT, phenotypic DST in the MGIT system and pncA sequencing. For more details, see the WHO Technical manual for drug susceptibility testing of medicines used in the treatment of tuberculosis (20).
a. If PZA resistance is not detected, or if PZA DST is not available, continue therapy with the regimen designed based on the previous DST results.
b. If PZA resistance is detected, individualized treatment regimens may have to be designed, especially if resistance to both FQ and PZA is detected.
4. If the INH result is uninterpretable or invalid, repeat the low/moderate complexity automated NAAT or FL-LPA with a fresh specimen. Consider conducting culture and molecular or phenotypic DST for INH on the isolate, if the patient is considered to be at risk of having Hr-TB.
5. For all patients, treatment monitoring should include collecting samples for culturing, as described in WHO guidelines. Any positive culture suggestive of treatment failure should undergo phenotypic and molecular DST, if available. At a minimum, DST should include testing for resistance to INH and RIF for patients on first-line regimens, and for RIF, FQs and PZA (if available) for patients on Hr-TB regimens. The treatment regimen should be modified as necessary, based on the results of the DST.
Interpretation of discordant results
This algorithm follows from Algorithm 1 with an mWRD that detected MTBC and was RIF-susceptible. In the scenario where the moderate complexity automated NAAT was used, the INH result would already be available. In this algorithm the follow-up testing could have a second RIF result when the FL-LPA is used as a follow-on or a second INH result when the moderate complexity automated NAAT is followed by the low complexity automated NAAT. Sometimes, results may be discordant. Each discordant result will need to be investigated on a case-by-case basis. General considerations are outlined below.
1. mWRD (e.g. Xpert Ultra) result "MTBC detected", follow-on FL-LPA "MTB not detected" or "uninterpretable".⁵⁸
a. mWRDs approved for detection of TB have a lower LoD than the FL-LPA; thus, FL-LPA may fail to detect TB in mWRD-positive samples that contain few bacilli. For example, it is estimated that about 80% of specimens with "MTBC detected" by Xpert MTB/RIF will generate an interpretable FL-LPA result.
b. The initial mWRD result should be used to guide treatment decisions, pending additional testing.
c. Follow-up actions may include submitting a specimen for culture and a molecular or phenotypic testing of the recovered isolate, and evaluating the possibility of laboratory or clerical error.
2. Initial mWRD result "MTBC detected, RIF resistance not detected"; RIF resistant by FL-LPA.
a. Treatment decisions should be based on the FL-LPA result (i.e. treat based on the worst-case scenario).
b. This result is expected to be rare because both assays interrogate the same region of the rpoB gene. There have been reports of mWRD RIF-susceptible and FL-LPA RIF-resistant discordances, but little data are available to assess how frequently this occurs.
c. FL-LPA is more sensitive for identifying RIF resistance than most mWRDs in hetero-resistant populations (mixtures of susceptible and resistant bacteria). The test includes hybridization probes specific to both the common mutated and the wild-type sequences in the bacterial genome. If the Xpert Ultra was used, a review of the probe melting temperature curves may be helpful to identify hetero-resistant populations (e.g. a double peak).
d. Follow-up actions may include DNA sequencing, conducting phenotypic DST and evaluating the possibility of laboratory or clerical error.
3. Moderate complexity automated NAAT result "MTBC detected, RIF resistance not detected, INH resistance detected"; but "INH susceptible" by low complexity automated NAAT.
a. This result is expected to be rare because both assays interrogate the same region of the katG and inhA genes.
b. The existence of hetero-resistant populations (i.e. mixtures of susceptible and resistant bacteria) is a more likely reason, especially in high-burden settings where the force of infection is high. A review of the low complexity automated NAAT probe melting temperatures (53) may identify such a possibility (e.g. dual peak).
c. Follow-up actions may include DNA sequencing, conducting phenotypic DST and evaluating the possibility of laboratory or clerical error.
d. Reassess risk for Hr-TB, and if high risk for Hr-TB or admin errors (e.g. mislabelling) are not the reason, treatment decisions should cover the worst-case scenario and be based on the moderate complexity automated NAAT result.
4. Moderate complexity automated NAAT result "MTBC detected, RIF resistance not detected, INH resistance not detected"; INH resistant by low complexity automated NAAT.
a. Treatment decisions should be based on the low complexity automated NAAT result (i.e. treat based on the worst-case scenario).
b. This result is expected to be rare because both assays interrogate the same region of the katG and inhA genes. However, the low complexity automated NAAT is more sensitive for INH detection as it includes additional gene targets (fabG1 and oxyR-ahpC intergenic regions).
c. The existence of hetero-resistant populations (i.e. mixtures of susceptible and resistant bacteria) is another possible reason, especially in high-burden settings where the force of infection is high. A review of the low complexity automated NAAT probe melting temperatures may identify such a possibility (e.g. dual peak).
d. Follow-up actions may include DNA sequencing, conducting phenotypic DST and evaluating the possibility of laboratory or clerical error.
For footnotes please see page 102
⁵⁸ The FL-LPA is not classified as an mWRD because it is not an initial test for TB detection but rather a follow-on test.