MSF reports experiences with the Gene Xpert MTB/Rif TB diagnosis assay
in different epidemic settings
Dr
Anne-Laure Page described the experience of MSF in implementing the
Xpert
MTB/Rif assay in routine conditions into four programmes with different
HIV
prevalence and risks of MDR-TB. The evaluation assessed per-patient
diagnostic
results, the effects upon treatment decisions and time to treatment and
feasibility (the number of tests per day, turn around time).7
The
four sites included Tbilisi in Abkhazia (low HIV, high MDR), Kampong Cham in
Cambodia (low HIV, low MDR), Mathare in Kenya, a slum area with high HIV, but
low MDR, and Nhlangano in Swaziland, an area with both high HIV and high rates
of MDR-TB.
The
diagnostic algorithms varied somewhat by setting: in high HIV settings, Xpert
and LED-fluorescence microscopy were the initial tests.
Xpert could be repeated
in patients who initially tested negative who did not respond to a course of
antibiotics. In the low HIV settings, smear microscopy was the initial test,
and Xpert was only used in smear-negative cases that did not respond to a
course of antibiotics, or in smear-positive cases at high MDR-TB risk.
Notably,
Kenya was the only site where the test was conducted on the same day as the
patient visit (in over 60% of the cases) — this was associated with a
significantly shorter time to treatment initiation.
“Same-day
testing is crucial to optimize delay to treatment initiation and decrease loss
to follow up,” said Dr Page.
In
high-HIV prevalence settings, there were varying relative gains in diagnoses
but it was consistently higher in HIV-positive rather than HIV-negative
patients.
“This
is in favour of using Xpert as the front-line TB test when HIV prevalence is
high. However, considering the possibility of an inconclusive result, a second
specimen remains useful,” said Dr Page.
In
low-HIV prevalence settings, Dr Page noted there was a lack of standardization
of Xpert MTB/RIF requests and algorithms were not strictly followed.
“As
a result of this as well as the limitations of routine data collection, there
was little information available on whether there is an added value of repeating
Xpert after a course of broad-spectrum antibiotics in subjects whose first
specimen is Xpert negative,” she said.
Two
concurrent sessions had further reports on the use of Xpert in a wide variety
of settings — and the extent to which it is increasing diagnoses and access to
treatment.
Some
of the key findings are described in the table below.
Patients’
descriptions:
|
HIGH HIV SETTINGS
|
LOW HIV SETTINGS
|
|
Kenya
(N=1047)
|
Swaziland
(N=2426)
|
Abkhazia
(N=734)
|
Cambodia
(N=1879)
|
Males (%)
|
58
|
44
|
71
|
52
|
Median age (years)
|
32
|
36
|
47
|
52
|
Tested for HIV (%)
- HIV-positive (%)
|
92
30
|
68
65
|
8
14
|
15
4
|
Tested with Xpert (%)
- Repetition (inconclusive) (%)
- Repetition (algorithm) (%)
|
98
12
2
|
99
3
0.2
|
40
5
NA
|
55
3
NA (9)
|
Rifampicin
resistance:
|
Kenya
|
Swaziland
|
Abkhazia
|
Cambodia
|
Proportion
rif resistance among Xpert-positive
|
5%
|
11%
|
19%
|
3%
|
Started on
MDR empiric treatment
|
7/9
|
25/30
|
15/17
|
3/3
|
Median delay
to treatment initiation (IQR), days
|
17
(6-17)
|
10
(7-25)
|
12.5
(7.5-28)
|
17
(10-25)
|
Costs of
diagnosis to health system
Some
of the presentations focused on how to reduce the costs associated with using
the test. In Cambodia, which has a low-HIV prevelance and very low MDR-TB burden,
Dr Kosuke Okada of the Research Institute of Tuberculosis
(RIT)/Japan Anti-Tuberculosis Association reported that diagnostic algorithms that
reserve Xpert as the third-line test (screening only smear-negative patients
without clear signs of TB on chest x-ray) would diagnose as many patients and
could save the TB programme up to 20-25% in costs over using Xpert as the
frontline test.8
Dr
Okada did not present data on time to treatment using these algorithms. Also
this approach will mean some people with MDR and people living with HIV might
fall through the cracks.
Xpert’s use in
settings with low access to chest x-ray
In
addition, these results won’t be the matched in countries with different
epidemiology. Even in India, a country where the HIV burden is low, “upfront
testing with Xpert can detect more pulmonary TB cases,” according to Dr Manoj
Toshniwal of the World Health Organization-Country Office for India.9
“But it is much more useful in the health facilities which do not have X-ray
facilities.”
In
India (and presumably in other countries as well), people with presumptive TB
who are smear-negative by microscopy must travel to a higher centre for a chest
x-ray to receive a diagnosis.
Following
WHO’s recommendation of the Xpert MTB/RIF, RNTCP, WHO and FIND collaborated on
pilot studies with Xpert in facilities at sites in India. Dr Toshniwal
described findings of introducing Xpert into one district in Amravati Municipal
Corporation comparing diagnoses at facilities before and after Xpert was introduced
in 2012.
A
similar number of suspects were screened each year, but in 2012, there was a
27% increase in pulmonary TB diagnosed — smear-negative diagnoses increased by
61%. The increase was even more substantial in the labs that did not have an x-ray
facility, where a 7-fold rise in case diagnoses was observed.
“Diagnosis
of pulmonary TB cases can be ensured by sputum specimen collection and
transportation to a laboratory having Xpert instead of referring the TB suspect
to the X-ray facility. The TB Control Programme should emphasise upfront
testing of TB suspects with such tests in district microscopy centres that do
not have X-ray facility. The availability of Xpert at district level improves
diagnosis of MDR-TB,” he concluded.
Introducing Xpert
where access to diagnostics is limited
Similarly,
a study in the Central Karoo, South Africa also demonstrated that introducing
Xpert substantially improved the diagnosis of TB and patient care in an area
with poor access to diagnostics and a high case fatality rate despite having a
relatively low HIV co-infection rate (around 11.4%) and low burden of MDR-TB.10
“We aimed to assess
the impact of Xpert implementation on patient care in the Central Karoo [in Laingsburg, Prince Albert, Murraysburg] – a rural, sparsely populated
area with poor access to
laboratory and radiology,” said Dr Theo Van Den
Handel of Right to Care.
The three health care facilities were selected
for their remoteness from laboratory and x-ray services, and a before and after
analysis of routinely collected data was performed to assess the impact on time
to treatment and proportion of cases with bacteriological confirmation.
Over
the course of 1020 observation days (half before and half post Xpert
implementation), 959 people were assessed (584 by smear microscopy and 375 by
Xpert).
“The
number of patients assessed for TB was 17%, 28% and 53% lower after Xpert
implementation, due to stricter adherence to the criteria for defining TB
suspects and we feel that the perception of higher cost played a role in this.
But despite the lower number of suspects assessed for TB, we still found the
number of TB cases diagnosed increased by 4%,” said Dr van den Handel.
There
was also a 12% increase in bacteriological confirmation, a 9% increase in
treatment initiation rate, and an 11-day reduction in time from first sputum
collection to initiation of TB treatment.
Xpert clearly improved the quality of diagnosis and health care in these
remote sites and also capacitated clinical staff in supporting treatment and
management decisions, according to Dr van den Handel.
“As far as the policy implications, when determining the optimal
placement and roll-out strategy for Xpert MTB/RIF countries may need to
consider characteristics of places — taking into consideration access to
diagnostics, diagnostic turn around time, poverty— in addition to the currently
recommended characteristics of people,” Dr van den Handel concluded.
Costs of
diagnosis to the patient
One potential reason for the improved access to diagnosis and
improved uptake of and time to treatment with Xpert in some contexts could be
the cost of diagnosis to the patient. According to Dr Anete Trajman, of Gama
Filho University, Rio de Janeiro, even though TB care is free-of-charge, there are
significant costs for patients including the cost of obtaining a TB diagnosis.11
She
reported the results of a study that compared the costs from the patients’
perspective of Xpert MTB/RIF as a replacement test versus that of the standard of
care (two smears) for the diagnosis of tuberculosis. The study was performed at
six primary care sites in Rio de Janeiro and 14 primary care sites in Manaus
from October 2012 – June 2013.
Adult
patients were interviewed up to four months after their diagnosis using a
standardised questionnaire, asking them about their sociodemographics,
out-of-pocket expenditures, food and transportation costs and time spent during
the diagnosis process, focusing on the bacteriological tests. Their indirect
costs were calculated based on the time invested multiplied by their hourly
income, based upon their current job activity. For those without formal work,
minimum wage was calculated at US$1.50 (around 3.5 Real) per hour.
218 patients were interviewed, with no significant difference between
those who received a smear microscopy diagnosis and those who received an Xpert
diagnosis — other than age (those receiving a smear diagnosis were slightly
older but age did not appear to effect relative costs).
All the costs were higher for those who received smear diagnosis — the
cost of transportation, hours lost per visit and total hours lost (smear
microscopy diagnosis took a median of one extra visit). Overall, the total cost
to the patient of receiving a diagnosis was a median of US $16.44 in the Xpert
arm versus $25.24 for those who received a diagnosis by microscopy — a
difference of $8.80.
“These costs are high, when you consider their income. Xpert was cost saving
in the Brazilian contexts, and we believe that in the future, with routine
implementation, Xpert can reduce one of the barriers in accessing TB
diagnosis,” said Dr Trajman.
Xpert MTB/RIF may miss some RIF resistance
in ‘mixed’ infections, leading to poor treatment outcomes
“Xpert
represents a major advance in TB diagnostics and has shown good performance for
the diagnosis of TB and RIF resistance in most settings,” according to Dr
Nicola Zetola of the Botswana-University of Pennsylvania Partnership.
Nevertheless, he believes clinicians should have a high index of suspicion for
MDR-TB in patients failing first-line therapy, despite an Xpert result showing
RIF-susceptibility.13
Dr
Zetola bases this on a study he performed in Botswana to see how sensitive
Xpert was at detecting whether someone has a mixed MTB infection. In high TB
(and HIV) burden countries, TB may be simultaneously caused by multiple
strains. In fact, mixed TB infections have been reported in 1% to 60% of TB
cases from endemic settings.
“We hypothesised that mixed TB infections (where at least one strain is rif-sensitive)
significantly decreases Xpert’s performance for the identification of
sub-populations of rif-resistant strains, leading to poor clinical outcomes,”
he said.
So
the study enrolled patients who had Xpert testing (using version 4 cartridge of
the test) at diagnosis and at least one culture with drug sensitivity testing
(processed using the proportion method in MGIT 960 media). Genotyping was then
performed to see whether the individual had a ‘mixed’ infection (note however,
that the mixed infection could indicate a mixture of drug sensitive MTB
strains, a mixture of drug resistant TB strains or a mixture of drug sensitive
and drug resistant strains.)
The study included 318 people diagnosed as being drug-sensitive TB by
Xpert, and 52 who were diagnosed as drug resistant.
There was a higher proportion of mixed infections than has previously
been reported in the literature: 20, or 6.3% among the Xpert ‘drug sensitive’
and 17 (or 32.7%) among the ‘drug resistant.’ Of these 37 mixed infections
taken together, 16 (43.5%) were a mixture of drug sensitive strains, 16 (43.5%)
were a mixture of drug resistant and drug sensitive strains, and 5 (13%) were a
mixture of drug resistant strains.
Roughly the same proportion of drug resistant and drug sensitive cured
or completed therapy — but deaths were more common among the Xpert sensitive
cases, with 19 (6%) dying, versus only 1 (1.9%) of the drug resistant cases
dying so far (some remain on treatment, and some are lost to follow-up).
Looking at whether Xpert and DST agreed on whether patients had mixed
infection, Dr Zetola said there was concordance in the results for mixed drug
sensitive, and mixed drug-resistant strains. However, Xpert missed four of
those which DST found to be a mixture of drug-resistant and drug-sensitive
strains. Xpert diagnosed these cases as being drug sensitive, and on the basis
of those results, the patients were started on first line TB treatment. All
four died.
Looked at overall, only some factors were associated with poor outcomes.
Being HIV positive (at any CD4 cell
count) — but especially with CD4 cell counts below 100 — is consistently
associated with poorer outcomes. But poor outcomes are also strongly associated
with having genotypic mixed infection (adjusted OR 6.5, 95% confidence interval
(CI) 2.1-20.5) and discordant results with DST (adjusted OR 6.6, 95% CI
1.2-48.2).
“Xpert
is not perfect. Given that mixed infections are increasing, clinicians should
be aware of the possibility of a false result in patients who are not doing
well despite having a sensitive test,” he concluded.
Xpert MTB/RIF
is getting more people with drug resistant TB onto treatment earlier — but
still not enough
While
Xpert MTB/RIF processes specimens very rapidly —and does shorten the time to
diagnosis and treatment—there were a number of presentations showing that
people (especially those with MDR-TB) are not getting on treatment as rapidly
as they should.
For instance,
Dr Riziki Kisonga reported that in Tanzania the use of molecular tests (including
both Xpert and the Hain line probe assay) cut the median time from MDR-TB diagnosis to start of treatment from 259 days to 34
days — but noted that the benefits were most pronounced among people living
with HIV and active TB.14
“Molecular
diagnostics should be scaled up to more areas focusing on the burden of TB/HIV,
he said.