A prospective multicentre study of discontinuing prophylaxis for opportunistic infections after effective antiretroviral therapy


Dr Phillip Hay, Department of GU Medicine, St George's Hospital, Blackshaw Road, London SW17 0QT, UK. Tel: +44 208 6721 255; fax: +44 208 6725 304; e-mail: phay@sghms.ac.uk



To assess the medium-term safety of discontinuing prophylaxis (primary or secondary) for opportunistic infections following an effective response to antiretroviral therapy.


Participating clinical sites prospectively identified patients in whom the discontinuation of prophylaxis for any opportunistic infection was considered to be clinically indicated, although CD4 levels were not predefined. A follow-up report was subsequently sent every 6 months requesting information on changes in prophylaxis, antiretroviral drugs, new AIDS-defining events, and CD4 cell count results.


Prophylaxis for Pneumocystis carinii pneumonia (PCP) was withdrawn in 524 patients (426 primary and 98 secondary prophylaxis), prophylaxis for Mycobacterium avium complex (MAC) was withdrawn in 28 patients (13 primary and 15 secondary), and prophylaxis for cytomegalovirus (CMV) retinitis was withdrawn in 10 patients. CD4 counts were generally maintained above accepted prophylaxis threshold levels during the period of follow up (95–98% of the time). Total follow up to last report or re-continuation of prophylaxis was 680 and 144 person-years for patients discontinuing primary and secondary PCP prophylaxis, respectively. No cases of PCP were reported, giving incidence rates of 0.0 (upper 95% confidence limit 0.4) and 0.0 (2.1) per 100 person-years. No cases of MAC were reported, but one patient had a recurrence of CMV retinitis. PCP prophylaxis was restarted in 30 patients; no patients restarted MAC or CMV prophylaxis.


Previous studies have demonstrated a low risk of PCP in the short term following the withdrawal of prophylaxis in patients who have responded well to antiretroviral therapy. The present study suggests a continuing low level of risk with extended follow up, provided adequate CD4 count levels are maintained. The case of recurrent CMV retinitis in a patient with impressive immunological and virological response indicates the need for close monitoring of patients previously diagnosed with this condition.


The incidence of opportunistic infections (OIs) in HIV-infected persons declined substantially with the widespread use of combination antiretroviral therapy [1]. This suggested the approach of discontinuing antibiotic prophylaxis in patients who experienced effective immune reconstitution, thereby reducing toxicity and drug interactions, lowering costs of care, and potentially improving adherence to antiretroviral regimens. A number of studies, both observational and randomized, confirmed the safety of discontinuing prophylaxis in patients whose CD4 cell count increased to levels indicative of a low risk of opportunistic infection [2–15]. Accordingly, the United States Public Health Service/Infectious Disease Society of America (USPHS/IDSA) guidelines were updated in 2002 [16], and specify the same or slightly more conservative CD4 count thresholds for discontinuing prophylaxis as those for initiating prophylaxis (Table 1). Although a large number of patients have been followed in relevant studies, the duration of follow up is short (typically less than 1 year) and data on the long-term safety of discontinuing prophylaxis are lacking. Here we report the results of the STOPIT study, a cohort study with mainly prospective follow up in which patients were followed for up to 4 years (median of 19 months) after discontinuing primary or secondary prophylaxis for an OI.

Table 1.   CD4 criteria for starting, discontinuing, and restarting opportunistic infection prophylaxis for adults with HIV infection [16]
Opportunistic infectionInitiating primary prophylaxisDiscontinuing prophylaxisRestarting prophylaxis
  • *

    Secondary prophylaxis.

Pneumocystiscarinii<200 cells/μL or>200 cells/μL for<200 cells/μL
pneumoniaoropharyngeal≥3 months 
Disseminated<50 cells/μLPrimary:Primary:
Mycobacterium avium >100 cells/μL for<50–100 cells/μL
complex ≥3 monthsSecondary:
  Secondary:<100 cells/μL
  e.g.≥6 months 
Cytomegalovirus >100-150 cells/μLCD4 count of
retinitis* sustained, e.g.<100–150 cells/μL
  ≥6 months 


In October 1998, all clinical centres in the Medical Research Council (MRC) Clinical Trials Unit network were invited to participate in a cohort study of HIV-infected patients who had discontinued prophylaxis (either primary or secondary) for an OI following an effective response to antiretroviral therapy. To reflect the variation in clinical practice at the time, ‘effective’ was not formally defined in terms of a specific CD4 count or viral load response but was left to the judgement of the individual clinician. However, patients were ineligible if prophylaxis had been stopped because of an adverse drug reaction or against medical advice.

Although most patients in the study were identified prospectively, patients who had discontinued prophylaxis within 2 years of the date of patient enrolment were also included; 81% of the total follow up was prospective. Information collected at study entry included a history of OI prophylaxis, the date and diagnosis of all AIDS-defining conditions, and, at the time of discontinuation of prophylaxis, specific antiretroviral drugs being taken and the two most recent CD4 and viral load measurements. Subsequently, a follow-up form was sent every 6 months requesting information on changes in OI prophylaxis and antiretroviral drugs, new AIDS-defining events, and CD4 cell count and viral load results.

This analysis includes data reported up to March 2002. Patients who were receiving prophylaxis for two OIs contribute to the analysis of each OI. Duration of follow up was measured from the date of discontinuation of OI prophylaxis to the earliest of the following: date of last clinical visit, date of resumption of prophylaxis, and date of diagnosis of the relevant OI.


A total of 612 patients from 37 centres were enrolled in the study. Sixty-four patients with less than 30 days follow up were excluded from the analysis. Of the 548 patients included in the analysis, 20 had discontinued prophylaxis for two different OIs [11 for Pneumocystis carinii pneumonia (PCP)+Mycobacterium avium complex (MAC), six for PCP+cytomegalovirus (CMV), two for PCP+toxoplasmosis, and one for MAC+CMV]. Prophylaxis had mainly been given against PCP (n=524), followed by MAC (n=28), CMV (n=10), toxoplasmosis (n=5) and cryptococcosis (n=1). Because of small numbers, incidence estimates were not calculated for the last two infections.

Characteristics of the study population at the time of discontinuation of prophylaxis are shown in Table 2. Demographic characteristics were similar in the three different OI groups, although patients who discontinued PCP primary prophylaxis generally had less advanced HIV infection than patients who discontinued primary prophylaxis for MAC; median CD4 cell counts were 354, 192, and 221 cells/μL, respectively. Protease inhibitor (PI)-based antiretroviral therapy (ART) regimens were slightly more common than non-nucleoside reverse transcriptase inhibitor (NNRTI)-based regimens.

Table 2.   Characteristics of the study population at the time of discontinuation of prophylaxis
  • PCP, Pneumocystis carinii pneumonia; MAC, Mycobacterium avium complex; CMV, cytomegalovirus; IQR, interquartile range; NRTI, nucleoside reverse transcriptase inhibitor; NNRTI, non-nucleoside reverse transcriptase inhibitor; PI, protease inhibitor.

  • *

    Some patients were in more than one transmission group; hence percentages do not add up to 100.

Median age (years) [IQR]38 [33–45]39 [33–46]40 [36–43]40 [34–43]43 [36–45]
Male sex [n (%)]334 (78)89 (91)11 (85)12 (80)8 (80)
Previous AIDS diagnosis [n (%)]108 (25)98 (100)8 (62)15 (100)10 (100)
Transmission group [n (%)]*
 Male homosexual250 (59)78 (80)6 (46)10 (67)8 (80)
 Heterosexual135 (32)17 (17)7 (54)4 (27)2 (20)
 Intravenous drug use17 (4)0 (0)0 (0)1 (7)0 (0)
 Other14 (3)1 (1)0 (0)0 (0)0 (0)
 Unknown19 (4)5 (5)1 (8)0 (0)0 (0)
Median duration of prophylaxis (months) [IQR]24.2 [13–41]28.9 [16–47]17.5 [11–28]29.3 [10–36]22.0 [10–34]
Median CD4 count (cells/μL) [IQR]357 [286–434]347 [290–447]180 [119–299]208 [141–274]221 [162–249]
% HIV-1 RNA <400 copies/mL8187466090
ART therapy [n (%)]
 NRTI+NNRTI161 (38)33 (34)5 (38)2 (13)3 (30)
 NRTI+PI194 (46)48 (49)6 (46)9 (60)7 (70)
 NRTI+NNRTI+PI29 (7)7 (7)1 (8)1 (7)0 (0)
 Other42 (10)10 (10)1 (8)3 (3)0 (0)

Of the 524 patients who discontinued PCP prophylaxis, 98 (19%) had previously been diagnosed with this infection, as had 15 of 28 patients (54%) who discontinued MAC prophylaxis. The median duration of primary prophylaxis before discontinuation was 24.2 months for PCP and 17.5 months for MAC; duration of secondary prophylaxis was longer, with medians of 28.9 months for PCP, 29.3 months for MAC and 22.0 months for CMV. The main drugs – either singly or in combination – used to prevent PCP were cotrimoxazole (80%), dapsone (10%) and pentamidine (9%), and those for MAC prophylaxis were ethambutol (54%), clarithromycin (46%) and azithromycin (39%). All but two patients received ganciclovir (80%) as CMV prophylaxis.

The decision to discontinue prophylaxis generally conformed to current clinical guidelines. Only seven patients (1.3%) who discontinued PCP prophylaxis had CD4 counts <200 cells/μL at the time of discontinuation; all CMV patients, and all but one MAC patient, had a CD4 cell count >100 cells/μL (Fig. 1). CD4 cell counts were generally maintained or increased during follow up, although 46 patients (8.8%) who discontinued PCP prophylaxis had one or more subsequent CD4 cell count measurements below 200 cells/μL. Notably, only 18 (39%) of these patients restarted PCP prophylaxis, as did 12 others who maintained CD4 levels above 200 cells/μL. Similarly, seven MAC patients and one CMV patient had a value below 100 cells/μL, although none resumed prophylaxis. CD4 values remained above the relevant threshold for 98, 95 and 95% of the total follow-up period for PCP, MAC and CMV, respectively (Fig. 2).

Figure 1.

 CD4 cell counts at discontinuation of prophylaxis for (a) Pneumocystis carinii pneumonia, (b) Mycobacterium avium complex and (c) cytomegalovirus.

Figure 2.

 Person-years of follow-up by CD4 count during the study after discontinuation of prophylaxis for (a) Pneumocystis carinii pneumonia, (b) Mycobacterium avium complex and (c) cytomegalovirus.

There were no confirmed or presumed cases of PCP following the discontinuation of primary or secondary prophylaxis over 680.1 and 143.5 person-years of follow up, respectively (Table 3). This yields upper 95% confidence limits of 0.4 and 2.1 per 100 person-years for PCP incidence. There were also no cases of MAC during follow up; upper 95% confidence limits for MAC incidence are 9.9 and 7.9 per 100 person-years following the discontinuation of primary and secondary prophylaxis, respectively.

Table 3.   Incidence of opportunistic infections following discontinuation of prophylaxis
InfectionNo. of patientsMedian follow-up
Total follow-up
No. of diagnosesIncidence* (upper 95%
confidence limit)
  • PCP, Pneumocystis carinii pneumonia; MAC, Mycobacterium avium complex; CMV, cytomegalovirus.

  • *

    Per 100 person-years.

  • Exact Poisson confidence limit.

PCP52419.1823.600 (0.4)
 Primary42619.7680.100 (0.4)
 Secondary9816.4143.500 (2.1)
MAC2830.567.600 (4.4)
 Primary1328.429.500 (9.9)
 Secondary1531.038.100 (7.9)
CMV1027.021.214.8 (22.6)

There was one presumptively diagnosed (by an ophthalmologist) case of CMV retinitis during follow up. CMV retinitis was previously diagnosed (laboratory confirmed) in August 1998 and following treatment the patient was placed on ganciclovir. This was discontinued 6 months later, in February 1999, when the patient's CD4 cell count reached 220 cells/μL on a regimen of lamivudine/stavudine/indinavir. There was a recrudescence of CMV retinitis in August 1999, at which time the patient had a CD4 count of 240 cells/μL and undetectable (<50 HIV-1 RNA copies/mL) viral load. The retinitis stabilized despite the patient declining treatment.

There were five deaths in the cohort, all unrelated to the OIs of interest, with causes of heart failure, cerebal haemorrhage, ascites, progressive multifocal leukoencephalopathy, and accidental death.


There is an extensive body of literature supporting the safety of discontinuing prophylaxis for opportunistic infections [2–15]. Most of the evidence relates to PCP prophylaxis, which was summarized in the meta-analysis by Trikalinos et al. [17]. This found an overall incidence rate of 2.3 per 1000 person-years among the 14 studies included. However, the follow up in most studies lasted between 0.5 and 1.0 years and provides reassurance only on the short-term safety of this approach.

This study, with mainly prospective follow up of up to 4 years, confirms the safety of discontinuing PCP prophylaxis over the longer term provided adequate CD4 count levels are maintained. Although discontinuation of prophylaxis was left to the judgement of the individual clinician, to reflect the variation observed in clinical practice, most decisions conform to current clinical guidelines. Only 1.3% of patients who discontinued PCP prophylaxis had CD4 counts of <200 cells/μL at the time of discontinuation. The study was not randomized, but such a design would not have improved the interpretability of the findings as no case of PCP was diagnosed. Also, the possibility of selection bias can be discounted since centres enrolled all eligible patients.

The number of patients who were observed to discontinue prophylaxis for MAC or CMV was small and definitive conclusions cannot be drawn from this study alone. Several larger studies have also concluded that withdrawing prophylaxis for MAC and CMV is safe [7–15]. However, one case of CMV retinitis was diagnosed in this study, 6 months after the discontinuation of ganciclovir, despite impressive immunological and virological response. Clinicians caring for such patients should be aware of the possibility of relapse or flare-up, which may be related to immune reconstitution syndrome.

There are currently no published data regarding levels at which prophylaxis should be restarted if, in the context of failing therapy, the CD4 count falls to levels at which the patient is likely to be at risk of OIs. In the absence of data, guidelines currently specify the same or similar thresholds as those for initiating prophylaxis. Our study provides limited evidence on this issue, with a total of 46 patients who discontinued PCP prophylaxis subsequently experiencing a CD4 cell count under 200 cells/μL. Despite the fact that prophylaxis was only re-started in 18 of these patients, no case of PCP was observed. With the efficacy of antiretroviral therapy proving to be durable, relevant evidence is likely to accumulate relatively slowly.


Appendix: study organization

The study was coordinated by the MRC Clinical Trials Unit (A. G. Babiker, E. Brodnicki, D. T. Dunn, H. Green, P. Kelleher and S. McCormack).

The following clinical centres participated: Addenbrookes, Cambridge (C. Carne, F. Wilson), Baillie Street Health Centre, Rochdale (H. Lacey), Castle Hill Hospital, Cottingham (C. Valentine), Central Middlesex Hospital, London (G. Brook, L. McDonald), Charing Cross Hospital, London (K. McLean, H. O'Toole), Chesterfield & North Derbyshire Royal Hospital, Chesterfield (P. Faser), Coventry & Warwickshire Hospital, Coventry (S. Allan), Doncaster Royal Infirmary, Doncaster (T. Moss), Farnham Road Hospital, Guildford (L. Howard), Glasgow Royal Infirmary/Gartnavel General Hospital, Glasgow (R. Nandwani), Kings College Hospital, London (C. Taylor, D. Graham), Leeds General Infirmary, Leeds (E. Monteiro), Leicester Royal Infirmary, Leicester (M. Wiselka, S. Bonnington, J. Laurenti), Mayday Hospital, London (A. Newell, M. Rodgers), Mortimer Market Centre, London (I. Williams, P. Newton), Newcastle General Hospital, Newcastle (P. Watson), Queen Mary's Hospital, London (P. Lister), Radcliffe Infirmary, Oxford (B. Morgan), Royal Berkshire Hospital, Reading (A. Tang), Royal Bolton Hospital, Bolton (E. Morgan, C. Bailey), Royal Devon & Exeter Hospital, Exeter (I. Alexander), Royal Hallamshire Hospital, Sheffield (C. Bradbury, S. Herman), Royal Infirmary of Edinburgh, Edinburgh (A. McMillan), Royal London Hospital, London (C. Skinner, P. Davis), Royal Sussex County Hospital, Brighton (M. Fisher, N. Perry, J. McIntosh-Roffey), Royal Victoria Hospital, Belfast (R. Maw, S. McKernan), St. Bartholomew's Hospital, London (M. Murphy, G. Roper), St. George's Hospital, London (P. Hay), St. Mary's Hospital, London (J. Weber), St. Mary's Hospital, Portsmouth (J. Tobin, L. Tucker), St. Thomas's Hospital, London (B. Peters), Upton Park Hospital, Slough (S. Dawson), Victoria Hospital, Blackpool (P. Flegg), Watford General Hospital, Watford (P. Munday), Whipps Cross Hospital, London (R. Melville), Whittall Street Clinic, Birmingham (M. Shahmanesh, G. Gilleran), Wycombe General Hospital, High Wycombe (G. Luzzi).