Tuberculosis

Diagnosis of pulmonary tuberculosis

Features on history and examination

Night sweats have to be drenching (as in a wet bed) to mean much in hot and humid regions.

Patients with many different conditions complain of loss of weight – it is worth asking about specifics such as clothes fitting less well. Do not be put off by an apparently short history; some patients deny symptoms lasting more than a few days even when they clearly have florid tuberculosis.

When examining, feel carefully for adenopathy, particularly cervical nodes in the supraclavicular area, which if found, are nearly always relevant. Careful respiratory examination is worthwhile – you should feel comfortable drawing the features you find in a diagram (and comparing it with the subsequent chest XRay to hone your skills.) Cursory chest percussion and mindless chest contact with a stethoscope is not adequate – palpate, percuss, and listen properly. Weigh your patient.

Radiology

Classical upper lobe cavitating disease is still found, but atypical presentations in HIV positive patients are also common. Non-specific infiltrates in either lower lobe can be tuberculosis, but another suggestive pattern in HIV positive individuals is a RML infiltrate, sometimes with hilar adenopathy. Cavitation is rare in these forms.

Tuberculosis and the sputum PCR

• South African tuberculosis care was transformed by the arrival of the PCR test (GeneXpert®) for tuberculosis detection. The sensitivity is > 95%, depending on clinical circumstances, in smear positive, culture positive cases and about 65% if smear negative and culture positive.¹
• Microscopy for AFBs only detects about 30% of culture positive, HIV positive patients. PCR has the added advantage of checking for rifampicin resistance. (Remember that if this is found, it is important to send a separate specimen for line probe assay and culture to properly determine the type of resistance and screen out XDR.) Do not send multiple sputum PCRs, as the incremental yield after a negative good quality specimen is low.

Sputum yield is considerably improved by paying attention to the process. Saliva is unhelpful. If sputum is not forthcoming, consider one or more of the following:

Chest physiotherapy (often difficult to find a physiotherapist, so try yourself).

Nebulisation with hypertonic saline: 5 ml of 5% (hypertonic) saline. (note that this is a potent source of an AFB aerosol if patients do have active cavitating disease, so ideally should be done outside or in a well-ventilated separate room.)

Don’t forget about the option of needle aspiration of lymph nodes.

Consider post-tuberculous fibro-cavitatory lung disease where symptoms are usually due to new viral or pyogenic infections rather than to recurrence of tuberculosis.

Tuberculosis and the urinary LAM

In hospitalised HIV positive patients with low CD4 counts (<100), it can be difficult to confirm tuberculosis even if there is a high clinical suspicion. In this situation the urinary LAM may be helpful.

Management – tuberculosis.

Patients with a typical clinical picture and positive sputum on TB polymerase chain reaction (PCR) testing should be started on treatment with four drugs (rifampicin, INH, ethambutol and PZA – RHZE), then dropped to rifampicin and INH (RH) after two months, but carrying on for a further four months to give six months treatment in total.

Patients who are classified as retreatment (previously treated, interrupted treatment, but not rifampicin resistant) get RHZE for three months instead of two, and then get rifampicin, INH (RH) and ethambutol (E) for five months. Streptomycin is no longer part of re-treatment as its use is not associated with improved cure rates, and it may lead to aminoglycoside resistance in poorly adherent patients at risk of MDR TB.

Drug	Months of Rx		Number of tablets per day
	First Rx	Re-Rx	30–37kg	38–54kg	55–70kg	>70kg
RHZE	First 2	First 3	2	3	4	5
RH 150/75	Last 4	Last 5	2	3
RH 300/150	Last 4	Last 5			2	2

There is often a desire to start therapy in the absence of diagnostic certainty. There are no rules:

In a fulminantly ill patient start treatment promptly as it may be life-saving.

In an ambulant undistressed patient try to prove the diagnosis first, even if febrile.

Many patients are somewhere in between these two, and decisions have to be a little arbitrary. An apparent need to start therapy in the absence of results should be accompanied by careful thought about what the next action will be if the patient: 1. does improve dramatically, or 2. fails to improve. Patients who appear to respond dramatically within one or two days of starting TB treatment often don’t have TB.

Perspective – the patient who is not acutely ill and is sputum negative on PCR.

Before HIV became so prevalent, the management of such a patient would be to send sputum for TB culture and repeat the CXR in 4–6 weeks. If the patient was better, the CXR was better, and the sputum was negative, the patient was reassured and discharged. If the sputum was positive, the patient was treated more or less regardless of clinical status. If the sputum was negative but the patient or the CXR was worse and no convincing alternative diagnosis was available, another sputum would be sent for culture and the patient considered for treatment.

Bed pressures created by the AIDS epidemic have made this approach appear somewhat leisurely, and there is also the perception that patients with tuberculosis and HIV may deteriorate very rapidly, so waiting two to six weeks for sputum culture results may be dangerous.

One approach piloted in 280 patients in Hlabisa took individuals who were sputum negative but strongly suspected to have pulmonary tuberculosis and gave sequential five day courses of amoxicillin and then erythromycin. Those who did not respond were considered likely to have tuberculosis. (Criteria for response were cessation of cough or sputum production, apyrexial for 48 hours, or the clinician thought that there had been a significant clinical improvement including a clear improvement in respiratory and pulse rates.) The criterion “TB because failed antibiotic trial” had a sensitivity of only 55% (spec 84%, LR= 3.4, LR- 0.5.)

(The specificity reported in Table 3 of the paper does not follow from data shown and has been re-calculated from the information in the text and tables)

Pre–test probability	0.01	0.05	0.1	0.25	0.5	0.75	0.9	0.95	0.99
Revised probability of TB
Antibiotic responsive	0.03	0.15	0.27	0.53	0.77	0.91	0.97	0.98	1.00
Not antibiotic response	0.01	0.03	0.05	0.14	0.33	0.60	0.82	0.90	0.98

The low sensitivity is of considerable concern, particularly when it is remembered that the gold standard in this case was itself a little shaky – not every patient with pulmonary tuberculosis has a positive sputum culture. In general then, in a patient who fails to respond to antibiotics and in whom your pre-test probability of tuberculosis is more than 75%, you can be considerably more sure that that is the problem; however if you are more than 10% sure the patient has tuberculosis, the finding of a positive response to antibiotics is NOT very good at excluding TB, and you should either think of other ways of investigating the patient, or consider starting on treatment if the patient is very sick, pending the culture result.

Perspectives – outcomes in smear negative tuberculosis.

A study form Malawi with a 20 month follow-up after treatment completion looked at this in a cohort of 248 patients (150 with smear negative pulmonary tuberculosis, and 98 with pleural tuberculosis), and of whom overall 83% were HIV positive. The mortality was 6.6 per 100 person years for those who were HIV negative, and 32.3 per 100 person years for those who were HIV positive. Recurrence was 6.8 per 100 for the former and 19.8 for the latter.

The mortality in the cohort as a whole was 25.7 per 100 person-years, which is an alarmingly high figure and acts as a reminder that HIV positive persons presenting with tuberculosis but negative sputum probably have fairly advanced immunosuppression.

Pleural tuberculosis

Pleural tuberculosis in HIV negative patients resolves without treatment in about two thirds of patients, but is followed by parenchymal or other extra-pulmonary disease in the majority within five years, and thus warrants prompt treatment. In high prevalence areas in patients with typical symptoms and signs, it is reasonable to treat based on characteristics of the aspirated fluid; the most important reason for aspiration is to ensure you are not dealing with an empyema.

Treatment is usually with standard therapy. Although success with fewer agents and shorter courses has been described, having a single standardized way of treating TB reduces confusion. Conventionally, the fever resolves within 2 weeks and the effusion within 6 weeks.)) For large effusions, a more reasonable expectation is 50% reduction in size at 2 months, 70% at six months, and 95% by a year. There is little advantage to mechanical drainage other than for very large effusions causing respiratory distress, and corticosteroids are of unclear benefit regardless of HIV status.

Managing the treatment environment (ward ventilation)

Remember that tuberculosis is an infectious disease, and that immunosuppressed individuals are particularly at risk if confined together in an enclosed space. (Lifetime risk of HIV negative individual contracting TB is about 10% in South Africa; yearly risk is 10% if HIV positive.) The catastrophic spread of XDR TB amongst AIDS patients in hospitals can be minimised by attention to some basic concepts.

AFBs are sensitive to UV light (sunlight.) However artificial UV light equipment has limited evidence of efficacy.

Ventilation provided by traditional high-ceilinged wards with open windows is efficient (and perhaps better than poorly maintained laminar flow.)

Ideal ventilation rates are about 12 exchanges of all the air in an isolation room each hour. This means completely fresh air everywhere every five minutes, and results in a perceptible breeze if working properly.

Patients who aren’t coughing aren’t particularly infectious.

Inducing sputum should ideally be done outside the ward and away from other patients or staff.

Be sensible about aerosol pracautions yourself – a snug mask (not a surgical mask) worn while interviewing or examining a patient with known or suspected tuberculosis is a very reasonable precaution, particularly as the incidence of MDR and XDR tuberculosis starts to rise.

Failure to respond to tuberculosis treatment.

Consider:

Wrong diagnosis

Another concurrent underlying illness (cancer, diabetes, SLE, HIV)

Not taking the treatment (the commonest), or treatment not well absorbed.

Inappropriate way of measuring response.

Drug resistant disease (MDR or XDR)

**Directly observed treatment (DOTS) and adherence

Overview –DOTS.**

DOTS programmes work, but the DOTS strategy on its own (without the programme involving infra-structural support, drug supply and administrative motivation) probably does not, in spite of the hype surrounding the process,.

Humans take medicines when they feel sick, and stop when they feel well.

If a pill makes you feel sick, you will struggle to take it.

Compliance is not always within the patient’s control, so think carefully before you blame somebody who genuinely couldn’t have done what you asked.

Combination therapy does not reduce the number of pills, but rather guards against MDR development by making the patient take either all or none of the therapy. If the treatment makes patients feel sick, some will rather stop.

Predictors of poor compliance are age (young), gender (males), HIV status (positive less likely to finish therapy) and work status (migrants struggle most).

Management – adherence issues.

Remember to notify the patient so that when, in spite of your best efforts, the patient is non-adherent, there may be a safety net of TB controller follow-up.

Make eye contact when talking about the importance of adherence.

Explain the rationale for continuing treatment when feeling better.

Tell the patient to report to your elected caregiver or yourself if the drugs cause nausea or vomiting, rather than just stopping treatment.

If there is a relative around, talk to him or her as well, preferably at the same time (it gives a ‘witness’ to the process, as well as saving time for yourself).

Organise a DOTS assistant. If you need to follow the patient personally (this should be rare!) specify follow up times, and make a plan so that you will know if the patient fails to return, so that you can then organise that the patient be traced. More usually, refer the patient to a clinic. Your letter should be legible and clear. Say if the patient has diabetes or epilepsy, and list the treatment (include doses!)

Ensure that you have slotted your patient into a system with adequate medical supervision, so that someone can say when treatment can be stopped.

Perspective – DOTS internationally

The success rate of tuberculosis treatment was over 90% in China and Vietnam, 80% in Ethiopia, and 66% in South Africa.

MDR tuberculosis

Management is hampered by variations in definitions and a paucity of good quality trials. Most of the efficacy information comes from small cohorts, and except for the newer agents where this is a registered indication, randomised controlled data is scarce. There is even less controlled trial information on multi-drug regimens. One review described the evidence base as a ‘dog’s breakfast’.

MDT TB is defined as in vitro resistance to INH and rifampicin with clinical non-response to conventional therapy. Extensively drug resistant (XDR) tuberculosis is MDR TB that is also resistant to a quinolone and an injectable (i.e. amikacin, kanamycin or capreomycin), and is a particular problem in immunosuppressed patients. (XDR is also abbreviated as EDR.) Patients who remain smear positive on first line agents should not be automatically assumed to have MDR TB: this diagnosis requires culture identification of in vitro resistance as well as failure of clinical response.

Management:

• Try to identify the cause. This is usually related to previous adherence issues, although de-novo resistance is well described. Adequate ‘adherence diagnosis’ and adherence intervention is essential, as in real terms fully supervised therapy for the ideal of 18–24 months doesn’t always happen.
• Beware of single isolations of commensal MOTTs (mycobacteria other than tuberculosis) in damaged lungs, which can cause confusion and which usually are not pathogens in the non-immunocompromised patient.
• All patients need to be managed, at least initially, in a facility designed to look after MDR TB patients, as access to medication is easier, Patients should not return to the community until they are sputum culture negative. Do not add single agents to a failing regimen. If in any doubt about what you are doing, consult a specialist of a colleague experienced in the use of these regimens. The intensive phase is four months, followed by a continuation phase of 5 months, bringing total duration to nine months.
•  

Linezolid (LZD) For the first two months 600 mg daily

Isoniazid (INH) high dose for four months: 10 mg/kg/day to max 900 mg/d (100 and 300 mg tablets)

Bedaquiline (BDQ) For the first six months. 400 mg/d for 14 days, then 200 mg Mon/Wed/Fri (i.e. 3 doses per week) (100 mg tablet)

Levofloxacin (LFX) 15 to 20 mg/kg/day. Max 1 g/d (250 and 500 mg tablets)

Clofazamine (CFZ) 100 mg/d

Pyrazinamide (Z) 30 mg/kg/day (usually 1.5 to 2 g/d) (500 mg tablet)

Ethambutol (E) 15 to 20 mg/kg/day, usually 800 mg or 1,2 g. (400mg tablet)

LFX, CFZ, Z and E are all given for the full nine months.

XDR TB treatment must also be given under supervision in a designated facility, and will usually involve therapy with a combination of at least five medications (including capreomycin 1g IMI daily (max 20 mg/kg/day), PAS (para-aminosalicylic acid), and terizidone 750 mg/d (max 20 mg/d) – e.g. 250 mg mane and 500 mg nocte.) The line between MDR and XDR treatment is now more blurred with the availability off linezolid and bedaqualine. Discuss with an expert.

As illustrated below, the basic principle is to have at least 5 agents to which the organisms is known, or is likely to be, sensitive. Move from right to left till you have selected this number.

Side-effects of second-line agents

Quinolones – GIT disturbances, headache, mood changes and, rarely, fits. CNS effects perhaps more of an issue with ofloxacin. Avoid in pregnancy.

Linezolid

Bedqualine

Clofazamine

High dose INH

Perspective – what is the outcome of MDR-TB treated with conventional therapy

A review of outcome in 1148 patients from 6 countries with drug resistant tuberculosis given standard short course therapy found 353 with MDR TB: mortality 47%, death 9%, and defaulted, completed or , failed 33%. About 11% were transferred. A study from Hlabisa in Kwazulu-Natal found a 2.3% MDR-TB prevalence rate in 561 patients in 1996, and of the 20 with follow-up data, 50% had died, 25% still had active disease, and 25% appeared to have been cured. It is a little difficult to know quite what to make of this information as the population is not clearly defined – the HIV prevalence was changing, and the proportion of this cohort who were itinerant workers returning home because they were sick is unclear. What the study did clearly show for that area, however, was that there had been a six-fold increase in the rate of MDR TB over three years.

The newer bedaqualine/linezolid/quinolone regimen has not been fully tested as an entity in randomised controlled trials, although the components have. Observational studies provide evidence that outcomes are considerably improved over the older regimens, and may in fact approach those of conventional tuberculosis.

A regimen of linezolid, bedaqualine and pretonamid as backbone² was shown in this observational study to produce a favourable outcome at six months in 90% of patients.

Perspective – how effective is treatment for MDR TB

There are many small cohort studies; generalisability is a major issue. A study from Peru gives some perspective, although their HIV prevalence rate was low. In that study, 55% were cured, 5% died, only 7% defaulted, and in 33% there was no response. (These figures differ a little from those quoted in the article abstract (48%; 12%; 11%; 28%) because these number refer to the group as a whole (13% did not have MDR TB but were just re-treatment patients)

The estimated NNT was thus about 2, with the cost per patient being about $322, and the cost per DALY gained being $270. The regimen used was kanamycin (for 3/12), ciprofloxacin, ethionamide, pyrazinamide and ethambutol daily for 18 months.

A study from South Africa (1996) yielded a 27% cure and 48% mortality at five years.

Prevention of nosocomial transmission of XDR TB

A mathematical model estimated the efficacy of various interventions in reducing projected nosocomial XDR TB infection in a district hospital. It found that the single most effective intervention (able to prevent 57% of infections) would be single bed isolation wards. Five bed isolation wards would prevent 37%, and improvement in natural ventilation 33% although this last had a wide confidence interval (8 to 35%.) Provision of masks (N95 for staff and surgical for patients) would only prevent 5% (95% CI 2 to 10%) and education about enforcement of mask policy only 1% (0% to 2%). Masks plus reduced inpatient stay plus natural ventilation performed as well as five bed isolation wards.

Isoniazid mono-resistance*

Resistance to isoniazid alone without resistance to rifampicin is relatively common and the earlier WHO recommendation was for 2 months of rifampicin, pyrazinamide, and ethambutol with INH (i.e. RHZE) followed by 4 months of isoniazid, rifampicin and ethambutol (i.e. RH plus EMB) in areas without access to INH susceptibility testing but with high prevalence of resistance, and 9 months (6 months after sputum conversion) of rifampicin, pyrazinamide and ethambutol, with or without a quinolone, if able to identify INH mono-resistance. In practice, most cases of resistance are only discovered after the patient has already been on standard four drug therapy for 2 to 4 months, and under such circumstances continuing with rifampicin, ethambutol and PZA for 6 months after sputum conversion, may be appropriate.

Clear evidence based guidelines on this are hard to come by as there is little controlled trial evidence. It is important to check sputum culture regularly for the emergence of associated rifampicin resistance.

Difficulty tolerating treatment

Nausea and vomiting on TB treatment.

This is probably the commonest important adverse effect, particularly early in treatment. There is a continuum ranging from a little mild nausea an hour or so after taking the pills, to intractable vomiting precluding any benefit from the medication which regularly ends up in a bucket by the bed. Particularly in HIV positive individuals, intercurrent gastroenteritis can cause vomiting and diarrhoea which is not due to the TB medication. Gastroenteritis in HIV may also reduce absorption of medication, which is listed as a reason for poor response to TB therapy.

Management

• Admit unless mild, particularly if the patient is frail, or with intercurrent HIV.
• Stop all TB medication for a day or so until the vomiting has settled.
• Start on intravenous or intramuscular metoclopramide 10 mg 8 hourly, and continue for a day or so until vomiting has ceased and the patient is eating well.
• Try starting treatment (in full dose) again.
• If this is not tolerated and there is moderate urgency to get a very ill patient onto therapy, then start on streptomycin, and ethambutol. Ethionamide is sometimes considered a tempting addition because it is readily available and the patient hasn’t been on it; however it is quite likely to cause considerable gastric irritation.
• Increase the dose of metoclopramide and try again.
• Do not continue antacids once the patient has settled, as these may interfere with absorption of the TB medication.

Inability to swallow pills

Some patients really struggle to swallow large tablets. Try crushing them, although this is frowned upon by the purists and is actually quite difficult to do.

Peripheral neuropathy and other walking difficulties

Peripheral neuropathy due to INH is probably not common enough to warrant the addition of pyridoxine to standard TB treatment, where its presence might mesmerise a few patients into stopping the big pills and only taking the little pills… The addition of pyridoxine is clearly worthwhile if there is any evidence of neuropathy before starting medication (look for it!) or if neuropathic symptoms develop on treatment. Do not wait for objective signs to develop! INH neuropathy is generally a purely sensory condition; the finding of motor symptoms or signs should prompt a search for alternative causes. Treatment is with higher dose pyridoxine (e.g. 50 mg 3x/d) and not just with the prophylactic dose of 25 mg per day. (A rarity is pyridoxine neuropathy – patients thought to have INH neuropathy and given high doses of pyridoxine sometimes respond to withdrawal of the pyridoxine.)

There are some patients with tuberculosis who present with weak legs before starting treatment, and this is something that should be actively pursued – all too often the weakness is ascribed to the general debility of severe tuberculosis and it may be weeks to months before the patient is examined properly. Apart from the peripheral neuropathies associated with malnutrition, consider:

• Tuberculosis with co-existent HIV infection, and an HIV associated neuropathy.
• Vasculitis associated with tuberculosis – may present with a predominantly motor neuropathy or a mononeuritis multiplex. Most of the information is from isolated case reports rather than from a clear literature.
• There are also reports of Guillain-Barre in association with tuberculosis; whether this is aetiological or coincidental is unclear.

If there are upper motor neurone signs in the legs, consider:

• Neuromyelitis optica is an association between optic atrophy and myelopathy; it has been weakly associated with tuberculosis.
• Tuberculous myelopathy – upper motor neurone features usually with a sensory level. If there is considerable pain, consider an associated tuberculous arachnoiditis.

Hepatitis

See drug induced hepatitis. In patients with severe chronic liver disease, consideration should be given to a regimen free of pyrazinamide, although the evidence for this is scanty. One option might be two months of streptomycin, rifampicin, isoniazid and ethambutol followed by six months of isoniazid and rifampicin. An even more conservative alternative would be two months of streptomycin, isoniazid and ethambutol followed by ten months of isoniazid and ethambutol. If in doubt, discuss with an expert.

Renal failure in patients needing TB treatment

Patients may have pre-renal uraemia secondary to vomiting or diarrhoea. If there is non-reversible chronic renal impairment, giving streptomycin or ethambutol is inadvisable and should only be done with extreme caution. Streptomycin in reduced dosage may be possible if renal function can be monitored carefully. (If GFR is less than 50, give the normal streptomycin dose every third day, if GFR is less than 10, give it every fourth day.) If ethambutol has to be given, reduce the mg/kg dose by 50% if the GFR is less than 50. However, a safer regimen is to give two months of rifampicin, isoniazid and pyrazinamide, followed by the usual four months of rifampicin and isoniazid. Some authorities recommend reducing the dose of pyrazinamide as well (give it alternate days) if the GFR is less than 10. If in doubt, look it up or ask!

Red urine and red tears

It is always worth telling patients that their urine and other body fluids may go red or orange on treatment – some individuals get so alarmed by this that they stop treatment.

Drug interactions

Anti-epileptics and TB treatment

Some texts state that TB medication can cause people on carbamazepine to develop signs of toxicity, but it is also common to encounter individuals on phenytoin who fit because of sub-therapeutic phenytoin levels. Be aware of these potential interactions.

Warfarin and TB treatment

Patients on warfarin who need TB treatment should have their INRs watched carefully; predicting what is going to happen is not always easy, but it is common to need to increase the warfarin dose considerably as the rifampicin induced hepatic enzyme induction kicks in. Similarly, individuals on TB treatment who require warfarin for the treatment of a DVT may require doses 2 or 3 times higher than usual, but the only way of finding out is to monitor carefully while initiating treatment.

ARVs

In patients on a dolutegravir containing first line treatment give an extra dose of dolutegravir (50 mg at night) while on rifampicin.

Bronchopleural fistulas in the context of active tuberculosis

The finding of a pneumothorax or an empyema in a patient suspected of having or known to have active tuberculosis raises some concerns. The conventional initial management of both conditions is to insert a chest drain. Unfortunately, quite a high proportion of such patients end up with a bronchopleural fistula, where the drain bubbles for weeks, often in the face of a relatively well if somewhat frustrated patient.

The management problem arises because of the difficulty of operating on patients with active pulmonary tuberculosis – poor healing and suture line dehiscence are common, particularly with lobectomies. A more aggressive strategy of pneumonectomy is also not always successful as the bronchial stump can break down.

Expect a more prolonged course of closed tube drainage, and refer back to a thoracic surgeon, if available, for review after about two months of anti-tuberculosis medication. Much earlier referral is ideal, but often difficult to achieve in situations where thoracic surgeons are difficult to access.

Modified regimens in patients unable to tolerate a component.

If in any doubt, consult an expert – it is remarkably easy to generate resistance!

Omitted drug		Suggested revised regimen
Rifampicin	INH, MOX, EMB for 16/12		Add aminoglycoside for 1st 2/12
Pyrazinamide	RIF, INH, EMB for 9/12

Unusual presentations of tuberculosis

Haematological phenomena.

The following are described: leukopenia, thrombocytopenia, leukocytosis (including leukaemoid reaction, where the WCC may be as high as 60 000), leukoerythroblastic anaemia, microcytic anaemia and polycythaemia. (Leukaemoid reactions occur in several other conditions – septicaemias, drug reactions, alcoholic hepatitis and the paraneoplastic syndrome. Differentiation from CML and other blood dyscrasias is by the clinical context, the benign smear, and a raised neutrophil alkaline phosphatase. The NAP is usually low in CML.)

Miliary tuberculosis.

This is a disease produced by blood stream spread, and more properly should simply be thought of as disseminated tuberculosis. Involvement of any organ, as mentioned below, can occur, but some patients have very subtle symptoms and signs – low-grade fever, malaise and weight loss as symptoms and then almost nothing to find on examination except pallor. The CXR classically shows ‘millet’ sized nodules (hence the name) but although these are usually 1 or 2 mm in diameter, they can be 5–10 mm, or absent entirely (in other words the CXR can be normal, even with hindsight). Diagnosis is based on the clinical picture and confirmed by finding mycobacteria in urine culture, sputum, bone marrow, or on liver biopsy. In ill patients with a typical CXR, commencing therapy in the absence of positive cultures is justifiable. (In a well patient, remember that pneumoconioses can give a similar CXR appearance.)

Lymph node tuberculosis.

Cervical nodes, particularly left supraclavicular nodes in HIV positive patients, have a high diagnostic yield for AFBs on needle aspiration. Another very common presentation of tuberculosis is with just mediastinal adenopathy on CXR in the context of an HIV positive individual with often relatively mild constitutional symptoms. Oesophageal symptoms occur rarely, and superior mediastinal syndrome has been described.

Soft tissue ‘cold’ abscesses in the neck are well recognised and are becoming increasingly common with the HIV epidemic. Less well recognised are indolent soft tissue swellings elsewhere which on aspiration contain pus. They can occur practically anywhere, although there is a suspicion that most are in some way related to the presence of a minor lymph node. Infraclavicular or other thoracic ones are sometimes thought to be related to pulmonary disease, but the so-called empyema necessitans, where a tuberculous empyema breaks through to the surface is rare.

Neurological

Meningitis.

Symptoms such as headache and malaise are classically considered to have been present for a month or two, but on close questioning some patients are adamant that they were entirely asymptomatic until a week or two previously. A whole range of neurological signs may be present ranging from mononeuritis through gaze palsies, cerebellar signs, chorea, and hemiplegias. Very rarely there may be meningitic inflammatory signs in the cord itself, with loss of reflexes, root pain syndromes and either flaccid or spastic paraplegia – the so-called ‘tuberculous arachnoiditis’

The CSF changes are classically a raised protein, low glucose, and lymphocyte predominance. In HIV positive individuals the changes may be less marked, with some indices normal. Some series have reported normal glucose concentrations (ratio to serum not < 0.5) in 5%. In both HIV positive and negative individuals the initial CSF may show a polymorphonuclear response rather than a lymphocytosis; repeating the CSF after 3 or 4 days is helpful under such circumstances. ‘Partially treated’ bacterial meningitis may be culture negative, but won’t usually show a dramatic shift in neutrophil/lymphocyte ratio. Do not neglect to send the CSF for TB culture.

The appropriate duration of therapy in TBM is not very well studied; common recommendations are at least nine months and possibly a year although one systematic review supported six months if the likelihood of resistance was low. Use RHZE for two months and then rifampicin and isoniazid for the remainder, all in standard doses. (Although ethionamide may have better CNS penetration, there is little clinical evidence of superiority over ethambutol, and it probably causes more GIT intolerance.)

CNS Tuberculomata.

These are essentially large CNS granulomas, and often present in patients on treatment for a month or so, perhaps for pulmonary tuberculosis. There is usually reduced consciousness, often with localising neurology and papilloedema. Diagnosis ideally should be made by the demonstration of the granulomatous lesions on CT scan, usually with considerable surrounding oedema. CT scan can show one or several lesions scattered in both grey and white matter, and varying in size. Often there may only be a single lesion but this is not sufficiently reliable to be used as a differentiator from toxoplasmosis, where the lesions are characteristically multiple. In the absence of access to a CT scan, the progressive neurological deterioration after 1–2 months of PTB treatment should make one strongly entertain the diagnosis, perhaps even sufficiently to start treatment with corticosteroids – e.g. betamethasone or dexamethasone 4 mg 8 hourly.

Sudden deterioration.

‘Sudden’ deterioration in individuals with neurological tuberculosis who are already on treatment should prompt consideration of the following:

• Hyponatraemia due to SIADH.
• Hydrocephalus.
• Uncontrolled TB – e.g. the TB Rx is not being given.
• Corticosteroids have been stopped abruptly.
• Stroke related to brainstem vasculitis (Examine the patient properly when admitted and then again if there is deterioration.)
• Undocumented epilepsy – ill patients with chronic meningitis unfortunately are often moved to side wards where their fits may not be observed.
• Hypoglycaemia.

Perspective – prognosis in tuberculous meningitis

The prognosis in tuberculous meningitis remains poor in developing countries, even with good quality TB regimens given for a full year. A study from Turkey found a mortality rate in non-immunosuppressed individuals of 27.8%, and the Vietnamese study quoted below had a mortality of 32% in the dexamethasone group and 42% in the control group.

Perspective – adenosine deaminase in CSF in suspected TBM.

This goes in and out of favour, but probably should remain out as there are too many false positives (crypto, lymphoma) and it isn’t very sensitive. A study using a ROC-generated cut-off of 8.5 IU/l in HIV positive individuals found a sensitivity of 57%, specificity 87%, and LR+ 4.4, LR- 0.5:

Pre–test probability		0.01	0.05	0.1	0.25	0.5	0.75	0.9	0.95	0.99
Revised probability
ADA positive		\0.04	\0.19	\0.33	\0.59	\0.81	\0.93	\0.98	\0.99	\1.00
ADA negative		\0.01	\0.03	\0.05	\0.14	\0.33	\0.60	\0.82	\0.90	\0.98

(i.e. when your clinical suspicion is high, a negative ADA is unhelpful.)

A systematic review came to the conclusion that the ADA cannot distinguish between bacterial and tuberculous meningitis, but then managed to reach the softer conclusions that tweaking cut-offs may be helpful: values less than 4 had a sensitivity of 93% for excluding TB (LR- 0.14), and values greater than 10 had a specificity of 91% but a sensitivity of only 49.5% (LR+ 4.72.)

Perspective – simple clinical features to improve diagnostic accuracy.

A study from Vietnam looked at age (>36 yrs), length of history (>6 days), blood total WCC (<15 000), CSF WCC (<900) and CSF neutrophil proportion (<75%). If all of these were present (‘test positive’), sensitivity for detection of TBM was 86% (spec 79%, LR+ 4.1, LR- 0.25.):

Pre–test probability	0.01	0.05	0.1	0.25	0.5	0.75	0.9	0.95	0.99
Revised probability
‘Test’ positive	\0.04	\0.18	\0.31	\0.58	\0.80	\0.92	\0.97	\0.99	\1.00
‘Test’ negative	\0.00	\0.01	\0.03	\0.08	\0.20	\0.43	\0.69	\0.83	\0.96

Perspective – doses of anti-tuberculous medication in neurological TB

Reduced CNS penetration has led to recommendations for the use of higher doses of anti-tuberculous medication in CNS tuberculosis. An open label Phase II trial form Indonesia using IV rifampicin 13 mg/kg, and oral moxifloxacin (800 mg/d), isoniazid, ethambutol and PZA found an impressive mortality reduction, but this needs to be confirmed.

Perspective – corticosteroids in tuberculous meningitis.

Several RCTs demonstrate benefit, but they nearly all have design flaws. A systematic review suggested reduced risk of death in children but not in adults, although only 158 adults were included. Subsequent to this, a larger trial (545 patients randomised) from Vietnam demonstrated no change in the primary composite outcome of death or severe disability, but a significant reduction in the risk of death with the use of IV dexamethasone for a month followed by tailing oral dexamethasone for a second month. If patients were fully conscious, they received only two weeks of IV therapy and then were given four weeks of oral treatment. The dose of dexamethasone was quite high – 0.3mg/kg/day IV for the first week (0.4 mg/kg/d if GCS 14 or less), 0.2 mg/kg/d IV for the second week, and 0.1 mg/kg/d PO for the third week. From the fourth week, the dose was 3mg/d, the 2mg/d in the fifth week, and 1mg/d in the sixth and final week. In patients with reduced consciousness, IV therapy was continued for a full month (0.4, 0.3, 0,2 and 0.1 mg/kg/d in each week) and the oral tail was for a further month (4m/g for a week, then 3 mg/d, then 2 mg/d, and then 1mg/d for the final week.)

The upshot of this was that there was a non-significant absolute risk reduction of 5.3% (95% CI -1.8 to +12.3%, NNT 19) for the composite primary endpoint of death or severe disability at nine months, although there was a significant mortality advantage of 9.6% (CI 2.8 to 16.3, NNT 11.) Hence, while more people were kept alive, more were severely disabled, leaving a non-significant trend to better outcome for the composite (and in this case clinically relevant) endpoint of death or severe disability. Subgroup analysis showed that benefit was significant for both death and (death+disability) only for patients who were fully conscious on admission.

Ophthalmological

A small yellow-white lesion at the limbus of the cornea, often with surrounding conjunctival injection, is known as a phlyctenule. Another lesion less often seen, perhaps because people don’t bother to look, is a retinal lesion called a choroidal tubercle that is a white blob about a quarter of the size of the disc. It can be single or multiple.

Cardiovascular disease in tuberculosis

Vessel wall tuberculosis (mycotic aneurysms)

Direct involvement of the vessel wall, either by contiguous spread from an adjacent node, or less commonly via the vase vasorum of larger vessels, or perhaps even through the endothelium, is well described but quite rare. The usual presentation is with mycotic aneurysm formation, and as this can be due to HIV itself, the differentiation can be difficult, and usually rests on imaging studies demonstrating the spread from a node, or on histological findings at resection. Mycotic aneurysms of the carotids are surprisingly often confused with lymph nodes, but FNAB of pulsating neck masses is inadvisable. Always palpate thoughtfully before sticking a needle into a lump in the neck (or anywhere else!) Although there is little work comparing medical and surgical approaches, in one series looking at aortic disease it was noted unsurprisingly that a combined approach using conventional indications for surgery was superior to simply treating the TB. Other even less common infectious causes (e.g. salmonella gastroenteritis) of false aneurysms are also described.

Pericardial tuberculosis.

See the section under Cardiology.

Skin involvement in tuberculosis.

An ulcer with associated adenopathy is known as a primary cutaneous complex. It is like a Ghon focus of the skin.

Lupus vulgaris consists of reddish papules on the skin that can be confused with Kaposi’s or psoriasis.

Tuberculids are less common and are thought to represent an immune phenomenon, although tuberculous DNA can sometimes be isolated from the lesions. Erythema nodosum is one manifestation; another is the so-called papulonecrotic tuberculid.

Scrofuloderma (colliquative tuberculosis) occurs when the skin is involved overlying a breaking down lymph node, but can also occur over a joint or bone focus. It has even been described as a rare but striking cause of proptosis. A similar phenomenon of direct skin involvement with tuberculosis, called orificial TB, involves the mouth, nose and/or anus in individuals with very severe pulmonary or gastrointestinal tuberculosis.

Verrucous tuberculosis consists of warty plaques on the dorsum of the hands in health workers involved with TB infected material (usually laboratory staff).

Hepatic involvement in tuberculosis.

Liver involvement in miliary tuberculosis is common, although marked hepatomegaly is not. It isn’t usually a primary way of presentation for tuberculosis, but sometimes can be a puzzlingly prominent feature of an otherwise poorly characterised illness of low-grade fever and weight loss. The LFTs may show a predominantly canalicular pattern (raised GGT and ALP). Biopsy showing AFBs is diagnostic, but if only granulomata are seen, some caution should be exercised as the liver reacts to many illnesses (sometimes including lymphoma) with granuloma formation.

Enteric/peritoneal tuberculosis.

Patient with peritoneal tuberculosis present with a constellation of rather non-specific gastrointestinal complaints such as pain, diarrhoea, swelling (ascites), and loss of appetite. The underlying feature of all these somewhat odd presentations is that the patient has features of systemic illness with fever, night sweats and weight loss. A doughy abdomen was only found in 9% in one series, and its specificity is unknown. In HIV negative patients aids to diagnosis come from a suggestive CXR (36–82%), associated adenopathy (rare) and the ascitic fluid. An ADA of more than 32 U/l in the fluid in a typical clinical context (high TB prevalence area and nothing else to suggest malignancy) makes the diagnosis very likely. Peritoneal biopsy, sigmoidoscopy, and bone marrow biopsy are seldom needed. Barium contrast studies may show a variety of lesions but seldom make a diagnosis. Therapy is the same as for pulmonary tuberculosis – there is no evidence that treatment for longer than 6 months confers any survival advantage.

Patients who are HIV positive can present with ascites and the rather non-specific doughy abdomen beloved by older texts. An alternative is with either no direct abdominal symptoms, or just abdominal pain. Ultrasound reveals retroperitoneal adenopathy, sometimes with central necrosis of the nodes, and often with some ascites. The spleen may also show multiple hypodensities. This constellation of findings is so specific that some would argue it is sufficient to make the diagnosis.

A variant of this is enteric tuberculosis that may be mistaken for appendicitis by the unwary (fever and right iliac fossa pain) although on closer questioning the symptoms have usually been present for weeks or months. It may also mimic intestinal obstruction, and it is not uncommon for the diagnosis to be made at laparotomy, where the characteristic thickened small bowel wall and associated adenopathy should ensure that a biopsy is sent for histology. A further trap for the unwary is pancreatic tuberculosis, which can present with clinical and ultrasound features suggestive of pancreatic malignancy. Unless tuberculosis can be demonstrated elsewhere, ultrasound guided transcutaneous fine needle biopsy of the pancreas is probably the procedure of choice.

Urinary tract tuberculosis.

Collecting system disease is easily recognised, but tuberculosis can present as a diffuse interstitial nephritis (not to be confused with the interstitial nephritis secondary to rifampicin therapy). Collecting system disease is diagnosed by ultrasound, culture of early morning urine specimens, or cystoscopy with biopsy. Menstrual irregularities or amenorrhoea and ‘PID’ are also described. The finding of positive urinary LAMs in many severely immunocompromised patients with disseminated tuberculosis is interesting, and raises the question of whether this is true renal disease, or glycoprotein leak from other areas.

Musculoskeletal tuberculosis.

Specific single joint arthritis due to tuberculosis is well described. Much rarer is acute polyarthritis associated with mediastinal tuberculous lymphadenopathy, and a mild self-limiting non-deforming arthritis called Poncets arthritis.

Spine.

Tuberculosis of the spine involves the disc with spread to the anterior aspects of the vertebral bodies above and below, resulting in vertebral erosions with loss of disc height. There is usually localised tenderness. Look carefully for the double shadow of a paraspinal abscess. (Metastatic malignancy tends to spare the disc, so one sees vertebral destruction with a preserved intervertebral disc height, and sometimes also loss of the pedicles.) The thoracic and lumbar spine are most frequently involved in tuberculosis, although cervical spine disease is described. Patients without neurological signs may respond nicely just to medication, but once there is any evidence to suggest cord compression, surgical intervention should be considered promptly, as progression to complete (and irreversible) paraplegia can be very rapid.

Other bones.

Other bony involvement can give an appearance similar to multiple myeloma, and tuberculosis should always be considered in the differential diagnosis of this condition, particularly since both can lead to high serum total protein estimations.

Perspective – adenosine deaminase in ascites.

Two articles, published at about the same time gave similar results. They were small (49 and 64 patients) but both gave sensitivity 100% and specificity 96%, yielding LR+ of 25.0 and LR- of 0.0. This yields the probability revisions shown below (ADA cut-off of 40 IU/ml):

Pre–test probability		0.01	0.05	0.1	0.25	0.5	0.75	0.9	0.95	0.99
Revised probability
ADA positive (ascites)		\0.17	\0.51	\0.69	\0.87	\0.95	\0.98	\0.99	\1.00	\1.00
ADA negative		\0.00	\0.00	\0.00	\0.00	\0.00	\0.00	\0.00	\0.00	\0.00

As a ‘rule out’ test, an ADA of <40 IU/ml appears to perform startlingly well, but it should be remembered that these were relatively early studies of suboptimal design by current standards and it is unlikely that this result would be reproduced in large high quality studies.

Perspective – how many sputum specimens?

In active pulmonary tuberculosis, the microscopy yield from sputum drops considerably as the number of specimens increases: the cumulative yield from two specimens is 92%, and 99% from three, with the incremental yield from the former being 15% and 8% for the latter.

Perspective – false positive sputum PCR after previous TB

There are case reports of sputum remaining positive by PCR up to five years after completion/cure of tuberculosis; consider this possibility particularly when there are reasonable alternative explanations for a patient’s symptoms (e.g. bronchiectasis exacerbation.)

Perspective – hypertonic saline sputum induction

A study from China found that using hypertonic saline nebulisation to induce sputum in patients suspected of having pulmonary tuberculosis but with ‘no’ sputum, inadequate sputum or sputum that was smear negative produced a quite impressive further yield: of 1648 such patients, sputum induction yielded AFB positive specimens in 558 (33.9% overall, 26.8% of 403 with no sputum, and 38.8% of 910 who were initially smear negative).

Perspective – TB PCR (GeneXpert®)

A Cochrane review pooled information from 18 studies:

	Sensitivity	Specificity	LR+	LR–
Initial test all patients	\0.88	\0.98	\44.00	\0.12
After negative AFB	\0.67	\0.98	\33.50	\0.34
Smear pos, culture pos	\0.98	\0.98	\49.00	\0.02
Smear neg, culture pos	\0.68	\0.98	\34.00	\0.33
HIV pos	\0.80	\0.97	\26.67	\0.21
HIV neg	\0.89	\0.99	\89.00	\0.11
To detect rif resistance	\0.94	\0.98	\47.00	\0.06

Perspective – doing more than one TB PCR

The SA guidelines suggest, at clinic level, one PCR, and if this is negative, sputum TB culture, a CXR, a trial of antibiotic, and (yes!) a clinical examination. There is some evidence to suggest that there may be some value in a second PCR – in one study sensitivity of a first PCR in smear neg culture positive disease in HIV positive patients was 43% for one specimen, and increased to 62% for two samples. The discrepancy between this result and the Cochrane review’s 68–80% sensitivity is that in the former study patients entering an ART programme were screened regardless of symptoms.

---

1. Steingart KR, Sohn H, Schiller I, Kloda LA, Boehme CC, Pai M, Dendukuri N. Xpert® MTB/RIF assay for pulmonary tuberculosis and rifampicin resistance in adults. Cochrane Database of Systematic Reviews 2013, Issue 1. Art. No.: CD009593. DOI:10.1002/14651858.CD009593.pub2. ↩

2. Conradie F, Diacon AH, Ngubane N, et al. Treatment of highly drug-resistant pulmonary tuberculosis. N Engl J Med 2020;382: 893–902. ↩

Was this article helpful?

1 out Of 5 Stars

1 rating

5 Stars		0%
4 Stars		0%
3 Stars		0%
2 Stars		0%
1 Stars		100%

5 Submit

Send Feedback

How can we improve this article?

Please submit the reason for your vote so that we can improve the article.

Submit