Approach to respiratory symptoms
The approach to respiratory disease in patients with HIV involves a synthesis of symptoms, signs, risk-factor history, knowledge of the patient’s degree of immunodeficiency, chest radiography and sputum sampling. More specialised investigations such as lung-function testing, computed tomography (CT) of the chest, radionuclide studies, bronchoscopy or lung biopsy may be required to make a diagnosis. Multiple disease processes may affect the respiratory system in human immunodeficiency virus (HIV) infection (Table 1) and multiple pathological processes can occur simultaneously, especially in advanced immunodeficiency. Non-infective respiratory disorders may also occur.
Opportunistic respiratory disease no longer predominates as the cause of respiratory symptoms in the era of combination antiretroviral therapy (cART). It is important therefore to consider the differential diagnosis of symptoms according to level of immunodeficiency and treatment with cART. The onset of respiratory symptoms within a few months of initiating cART in an immunocompromised patient suggests the possibility of immune reconstitution disease.
Knowledge of the level of immunodeficiency, as in many aspects of HIV disease, provides a guide to the likely differential diagnoses. People with significant immunodeficiency (e.g. CD4 T-lymphocyte (CD4) cell count < 200 cells/μL or < 14% of total lymphocyte count) are at increased risk of opportunistic infections, including Pneumocystis jirovecii pneumonia (PJP) (previously Pneumocystis carinii pneumonia [PCP]). Bacterial pneumonia can occur at any CD4 cell level, but occurs more frequently in immunodeficient people than in those with a preserved CD4 cell count, often in association with bacteraemia. Non-Hodgkin lymphoma and tuberculosis (TB) can occur at any level of immunodeficiency, although TB is more likely to involve extrapulmonary sites in immunodeficient patients. More severe immunodeficiency (CD4 cell count < 100 cells/μL) is associated with an increased risk of Toxoplasma gondii pneumonitis and cryptococcal pulmonary pathology. At this level of immunodeficiency, staphylococcal and gram-negative bacterial pneumonia occur more frequently than in people with CD4 cell counts > 100 cells/ μL. Infections with Streptococcus pneumoniae and Haemophilus spp. also occur. In the profoundly immunodeficient person (CD4 cell count < 50 cells/μL), both cytomegalovirus (CMV) and Mycobacterium avium complex may be associated with pneumonitis, usually in the context of disseminated disease. It is important to remember that all opportunistic infections may rarely occur at higher CD4 cell counts, especially in the setting of antiretroviral therapy.
Certain features in the patient’s history may be suggestive of one disease process over another. Typically, PJP is a subacute illness of fatigue, fever and dry cough of several weeks duration. In comparison, symptoms of short duration (hours to days), including chest pain, abrupt onset of fever and rigors and a productive cough are more consistent with bacterial (e.g. pneumococcal) pneumonia. Dyspnoea may occur in a range of infective pathological processes, but may also be the only symptom of cardiomyopathy, ischaemic heart disease or pulmonary hypertension.
Immunodeficient patients who are taking PJP prophylaxis are nine times less likely to develop PJP than those not taking PJP prophylaxis. Prophylactic treatment with cotrimoxazole (sulphamethoxazole/trimethoprim) is likely to reduce the incidence of bacterial infections as well as PJP and T. gondii reactivation. Persons with a positive Mantoux test (> 5 mm) who have not received treatment for latent TB have a 14% risk of developing active TB over a 2-year period and should therefore be considered for isoniazid prophylaxis. The data on gamma interferon release assays, such as Quantiferon Gold or ELISPOT testing, in the setting of HIV are more limited, but it is likely a positive test has the same significance. It is uncertain if there is a clinical benefit of isoniazid treatment for latent TB infection when defined by a positive gamma interferon release assay if the Mantoux test is less than 5 mm (see section on Mycobacterium tuberculosis).
People who have resided in or travelled to countries with a high prevalence of TB are at increased risk for disease. Travel to areas endemic for the dimorphic fungi may be important: Histoplasma capsulatum is endemic in areas of the USA (especially the Mississippi and Ohio River valleys), Latin America, West Indies, Eastern India, and Africa. Coccidioides immitis is hyperendemic in some parts of the southwest of the USA, Northern Mexico, and in Central and South America. Other illnesses with respiratory symptoms where a travel history might be important include melioidosis, penicilliosis, strongyloidiasis and Salmonella infection.
Smoking is an important risk factor for bacterial pneumonia, bronchitis, chronic airflow limitation and carcinoma of the lung in the population with HIV, as in the general population. Marijuana smoking has been associated with Aspergillus infection. A contact or occupational history may also be useful in suggesting the possibility of TB. A past history of incomplete treatment of TB may suggest drug-resistant disease. Current injecting drug use is a risk factor for infective endocarditis and staphylococcal pneumonia. Men who have sex with men are more likely to develop Kaposi sarcoma (KS), although isolated pulmonary disease without mucocutaneous lesions is uncommon. Antiretroviral treatment itself may be a cause of respiratory symptoms, e.g. dyspnoea caused by nucleoside analogue-induced lactic acidosis, hypersensitivity reaction to abacavir (now largely eliminated by HLAB57*01 testing) and the largely unexplained increased incidence of bacterial pneumonias associated with enfuvirtide use.
The physical examination may be normal in patients with HIV with PJP, aside from tachypnoea and a dry cough, although fine crackles may be present. Signs of consolidation suggest bacterial, mycobacterial or fungal disease. Examination for extrapulmonary features should be undertaken, as disseminated disease processes may present with pulmonary symptoms. Cardiovascular signs should be assessed as pulmonary hypertension, endocarditis, cardiomyopathy and pericardial effusions may also cause breathlessness. Signs associated with immunodeficiency should be elicited (e.g. oral candidiasis) as these may influence the differential diagnosis in the absence of a current CD4 cell count. The skin should be examined for the presence of KS and the cutaneous lesions of cryptococcosis. The presence of neurological signs may suggest disseminated toxoplasmosis, and drowsiness may occur in cryptococcosis. Hepatosplenomegaly may be present in people with Mycobacterium avium complex infection, non-Hodgkin lymphoma or CMV infection.
Most standard biochemical and haematological tests are not helpful in the diagnosis of respiratory symptoms in advanced HIV disease. However, peripheral blood neutrophilia and left shift may occur in bacterial pneumonia and cryptococcal disease. Neutropenia may suggest an increased likelihood of bacterial or fungal infection. Measurement of the degree of impairment of gas exchange with arterial blood gas analysis is important, as the degree of hypoxaemia may change the management of certain diseases (e.g. steroid therapy in PJP). In cases of suspected cryptococcal disease, the titre of serum cryptococcal antigen is used as evidence of disseminated disease. Bacterial blood cultures should be obtained as well as mycobacterial blood cultures in immunodeficient patients, given the elevated rates of bacteraemia.
The chest radiograph may reveal the presence of focal consolidation suggestive of bacterial, tuberculous, cryptococcal or lymphomatous disease. The typical radiological pattern of TB, with upper lobe disease and cavitation, may not be seen in immunodeficient people. Pleural effusions may be caused by KS or the range of diseases which also cause consolidation. The presence of a pleural effusion makes the likelihood of PJP quite low, although in advanced HIV disease, multiple disease processes may co-exist. The presence of a pneumothorax makes PJP more likely, although other disease processes may rarely cause a bronchopleural fistula and be associated with a pneumothorax. Cavitating disease is likely to be caused by Mycobacterium tuberculosis, bacteria, fungi, Nocardia spp., and some atypical mycobacteria. Diffuse infiltrates with a reticular or granular pattern are consistent with PJP, but many other diseases may have a similar radiographic pattern. A normal chest radiograph does not exclude PJP. Similarly, a normal radiograph may be seen in cases of TB or cryptococcosis where extrapulmonary disease is the major manifestation (see Table 2 and Table 3).
Sputum from all patients with HIV presenting with productive cough should be sent for microscopy and culture. Sputum investigations should routinely include Gram stain, bacterial and fungal culture and sensitivity testing. Although not all specimens need to be examined for mycobacteria, a low threshold for testing sputa for acid-fast bacilli and mycobacterial culture should be maintained. An induced sputum should be collected where the clinical likelihood of PJP is high, provided adequate negative-pressure facilities exist. The sample is assessed for the presence of P. jirovecii by immunofluorescence, Giemsa or silver stains and a result is usually available within 2 hours. More recently PCR testing for Pneumocystis has been developed, but the use of this technology is limited by poor specificity for disease. Induced sputum which is negative for P. jirovecii does not exclude the diagnosis of PJP. Supplemental investigations such as bronchoscopy increase the diagnostic yield for P. jirovecii, and may facilitate other diagnoses as well. P. jirovecii may still be detected by bronchoscopy and bronchoalveolar lavage after several days of therapy with cotrimoxazole. A high resolution CT scan that does not reveal ‘ground glass’ changes makes the diagnosis of PJP unlikely. Similarly, the pulmonary diffusing capacity is typically reduced and lung clearance of technetium is typically increased in PJP. The absence of these findings suggests that alternative pathology to PJP should be considered. Where a pleural effusion has been detected, a diagnostic pleurocentesis is warranted. Pulmonary nodules should be biopsied if sputum sampling, bronchoscopy and less invasive investigations are negative (e.g. serum cryptococcal antigen or bone marrow biopsy in the setting of pancytopenia).