Twenty-one years into the HIV pandemic the World Health Organization (WHO) has loosened itself from its slothlike lethargy and bureaucratic torpor to produce, seemingly with great reluctance and certainly with great confusion, a draft document on "Antiretroviral Guidelines for Resource Limited Settings."

In a justly scathing critique, Gregg Gonsalves wrote that at "the Interim WHO ARV Treatment Working Group in November 2001, it was clear that WHO was either deliberately as a matter of policy or inadvertently as a matter of disorganization, unwilling to unequivocally state its goals for AIDS care in the developing world: was it to support Ministries of Health and national AIDS programs to scale-up antiretroviral therapy and other AIDS treatment to treat large numbers of people or was it to support individual doctors and small clinics in managing individual patients on ARVs?"

Whichever its aim, the current WHO draft document fails. It lacks the clarity, simplicity, and overall strategy provided by programs such as STOP TB with its Directly Observed Therapy Short-Course (DOTS) and DOTS-Plus recommendations. It is obvious that HIV treatment does not lend itself to such simplicity, but the WHO draft document does not really try.

Nor does the WHO draft document provide much in the way of meaningful guidance to physicians for AIDS care. It would be more useful, perhaps, to simply ship thousands of copies of Bartlett and Gallant's Johns Hopkins Hospital Guide to Medical Management of HIV Infection to physicians in developing countries. Of course, the latter guide, while accurate and comprehensive, may not prove relevant in resource poor settings where many of the drugs, diagnostics, and medical infrastructure relied on as givens by Bartlett simply are not available.

But the WHO has not made an effort to develop a treatment scale-up strategy with anything like the effort shown over the past decade with STOP TB, DOTS, and DOTS-Plus. Whatever their limitations and failings, these programs have, at least according to the WHO, greatly expanded coverage in many countries, and substantially reduced TB morbidity and mortality.

As Gonsalves wrote, "the WHO guidelines in their final form are going to have to address one of the two goals addressed above. Right now, the document displays the contradictions of WHO's position on ARVs: it offers recommendations for scale-up and [recommendations for] individualized patient management, which frankly will be confusing to Ministries of Health [MOHs] and individual doctors... It is safer [for WHO] to offer guidance to MOHs that hedges saying if you'd like to scale-up you can do this, bu tif you don't want to do so, you can give these guidelines out to doctors in private practice in your country and [have them] use them for individual patient management." He makes several additional key points:

1. The WHO must make a commitment to help countries scale-up wide-spread access to antiretroviral treatment programs for HIV infected people whom they will benefit.
2. The WHO must provide regional and country-level technical support to ministries of health on how to ramp up broad access to antiretroviral treatment programs.
3. The WHO, with other multilateral agencies, foundations, and professional societies, should provide many different and synergistic or interacting forms of professional training and individualized guidance to physicians, nurses, and other health care workers appropriate for a wide variety of settings in which people with HIV will receive care.
4. The WHO lacks sufficient professional staff with broad-based experience treating HIV/AIDS in developing (or for that matter developed) countries.

   There are additional points to be made:

5. The entire structure of the agencies responsible for HIV/AIDS treatment and care in developing countries has been in flux, inscrutable, and indecipherable for several years. Recently UNAIDS underwent a much-vaunted "reorganization" which left things even more confusing than before. It appeared that most of their staff who knew anything about HIV treatment research and care moved to WHO, or elsewhere. But the structure of the WHO and UNAIDS programs lacks accountability, transparency, and many of the other desiderata of a well-run program. So far as I'm aware, between them the WHO and UNAIDS have fewer people of expertise working on HIV treatment than work in a single branch of the Division of AIDS at NIAID, NIH.
6. Because they are under-resourced, under-staffed, and under-led, the entire project over the last year to develop ART Guidelines for Resource-Poor Settings has been done in haste, on the fly, and in between other projects more pressing for the bureaucrats in charge.
7. In the meantime countries have been asked to submit projects to the equally under-resourced Global Fund for AIDS, TB, and Malaria, without any evident or coherent guidance from WHO or UNAIDS about what a good treatment program might look like.

As Gonsalves put it pithily, "One might expect the world's most important [?] institution devoted to public health to have a firmer grip on the question."

What happened to the ambitious vision articulated by Bernard Schwärtländer at the May 2001 WHO meeting on ART in RPS, where he articulated planning now for a WHO/UNAIDS guided strategy to scale up now to provide HAART to five million people in developing countries within five years? This goal is feasible and would be the minimum necessary to make a dent in AIDS related mortality in that time period. Of the world's 40 million HIV+ people, perhaps 10 million have low CD4 counts or AIDS and will need treatment in the next two years to survive. If we cannot treat even half of those needing therapy within five years we will failed on a massive scale.

In any case there are many additional points to be made about the draft guidelines themselves, which could be either a policy tool or a clinical management tool, but now fail as either.

After five years of diligent and often maddening participation in the US HHS Panel on Clinical Practices for the Treatment of HIV Infection, which has released the periodically updated Guidelines for the Use of Antiretroviral Agents in HIV-1 Infected Adults and Adolescents, I know how much the early form such guidelines take can affect, sometimes adversely, the health care decisions made by thousands of physicians for hundreds of thousands of patients. Certainly in the US the decision made in 1997 to recommend therapy for all HIV-infected individuals with fewer than 500 CD4 cells/mm³ or more than 5,000 copies of HIV RNA per cubic milliliter led to the premature initiation of therapy for as many as 65% of those who subsequently began HAART, thus leading many to develop, prematurely, adverse events, drug resistance, reduced quality of life, and increased cost of care.

Like sausage making, you wouldn't always want to know how Guidelines get made. It's a highly politicized process in which the scientific, ideological, or policy views and goals of the participants mix in an unpredictable way with ever-changing amounts of clinical data, laboratory data, gullibility, hope, fear, euphoria, greed, and ambition (among other things). Recommendations are based on an unstable and unpalatable combination of good clinical data, less reliable observational data, even less reliable and often capricious expert opinion, pragmatic concerns, and sheer bias or self-interest.

In the US, many forces combined to recommend the translation of "hit early, hit hard" from theory to practice guidelines. Some researchers had hopes of eradicating HIV. Most clinicians were appalled by the devastation of late-stage AIDS. No one knew whether immune restoration would be possible, let alone robust. Drug companies wanted to sell drugs. Policy-makers wanted to ensure that the drugs' costs were covered by the country's fragmented health care financing system. It is ironic but true that the staunchest critics of "hit early, hit hard" were the community representatives on the HHS Panel, and it was they who led the charge, from 1998-2002, to change the Guidelines to "hit hard, later."

Some Problems With the Draft WHO ART Guidelines

Throughout the document, there are no references. This is an important omission. The credibility of the document will rest at least in part on the quality of its references. The document introduction is poorly written. "Juxtaposition of events?" "Awareness of the scope of the problem?" Come on. "An opportunity to discuss and redress this circumstance?" A ringing call to action this is not.

The document should be a mobilizing force, not an argument with itself.

Throughout I comment on the draft document while providing suggestions in italics.

Introduction

This document is intended to provide guidance to governments and providers to initiate wide-spread scale up of antiretroviral treatment programs for people with HIV living in resource poor settings. The strategy for this scale up involves providing simplified antiretroviral regimens and strategies with potency and efficacy equal to standard of care used in the developed world, but in some cases with less laboratory monitoring when it is not immediately feasible everywhere. Nonetheless these guidelines provide recommendations for clinical and minimal laboratory management of HIV infected patients with highly active antiretroviral therapy.

Stronger emphasis is needed in paragraph three of the abundance of evidence from randomized and observational studies that highly active antiretroviral therapy (HAART) protects health and prolongs life when initiated and used among those with AIDS, symptomatic HIV infection, or fewer than 200 CD4 cells/mm³ (Abbott ritonavir study, ACTG 320, Palella NEJM 1998 HOPS data, French cohort, EuroSIDA, BC Cohort, etc.).

As a first principle, one would think that HAART would strongly be recommended for all people with AIDS, symptomatic HIV disease or, when available, CD4 counts measured below 200/mm³ in the absence of a non-AIDS infection or post-immunization.

Because of abundant evidence of HAART's clinical efficacy and ability to protect health and prolong life in this population, WHO strongly recommends that all people with AIDS, symptomatic HIV infection, and if measured, CD4 counts below 200 cells/mm³ receive HAART.

As a second point, either in the same or in a closely related document, diagnostic and treatment recommendations will be made for the management of HIV-associated opportunistic infections, cancers, co-infections such as tuberculosis and hepatitis, and palliative care.

Both HAART for people with AIDS and treatment for opportunistic complications would be part of a country's first widespread scale-up of AIDS care.

Section II. When to Start ART

Clarify, simplify:

1. All persons with an AIDS-defining condition* should be given HAART.
2. All persons with symptomatic HIV infection -- including [list symptoms] should be given HAART.
3. All persons with CD4 counts lower than 200 cells/mm³, where such tests are available, should be given HAART.
4. If CD4 cell counts are not available, all persons with total lymphocyte counts below 1200/mm³ should be considered as candidates for HAART.
5. If HIV RNA tests are not available, they are dispensable as CD4 count is the better predictor of the immediate risk of progression to AIDS.

   * A decision needs to be made regarding whether or not TB should be considered an AIDS-defining condition here or whether TB should be treated first for TB, given HAART at the same time or after clearing TB, or whether coinfected people should be stratified by CD4 counts and only those with low counts given HAART.

   Regarding earlier starting, the trend in most developed countries has been away from viral load based starting triggers, and towards lower CD4 based starting triggers now all starting between 200-350 CD4 cells/mm³. I see no need to make strong treatment regimens for people with above 200 CD4 cells since the evidence is that there is little harm in waiting to start therapy until that point. The key thing would be clinical monitoring for symptoms and laboratory monitoring when available for CD4 count drops.

6. HIV infected individuals should be in regular clinical care and followed for the development of symptoms, in which case therapy should be initiated.
7. When CD4 testing is available, asymptomatic HIV-infected persons should be tested at least every six months, and therapy initiated when the CD4 count drops below 200, even when symptoms are absent.

Section III. Recommended First-Line Regimens

This section is wordy and poorly-written. It introduces nuances which are completely irrelevant here, such as the distinction between nucleosides and nucleotide analogues (here dubbed NsRTIs and NtRTIs). Who cares? They're all DNA chain-terminating prodrugs which get phosphorylated twice or thrice and incorporated into newly synthesized viral DNA by RT. Why would we want anyone to have to remember this petty distinction when so many more important issues must be memorized and operationalized?

Clarify, simplify:

Based on current research data and clinical experience from a variety of randomized and observational studies carried out with the sixteen approved antiretrovirals over the past tow to ten years in tens of thousands of HIV-infected persons in the developed world, we recommend three possible therapeutic strategies:

A1. First line therapy with a dual NRTI, single NNRTI regimen
B1. First line therapy with a triple NRTI regimen

A2. Second line therapy with a new dual NRTI, single PI regimen
B2. Second line therapy with a new dual NRTI, single NNRTI or PI regimen (for B)

For the nucleoside backbone, to enable drug synergy and ease of use (no more than bid dosing), we recommend combinations of AZT/3TC, d4T/3TC, d4T/ddI, AZT/ddI, or ddI/3TC.

[Are these too complicated for Ministries of Health? There is an irreducible complexity to AIDS care. Simplifying things too far may raise failure rates, e.g., through toxicity and the inability to substitute a drug because regimens are so standardized that alternative, perhaps equipotent drugs are not available, and may actually reduce a program's effectiveness. On the other hand, putting everything out there without any guidance will lead to improper use of the drugs. Policy decisions have to be made which include the development both of guidelines for broad, rapid scale up with simplified HAART regimens, and more detailed guidelines for individualized care.]

[Question: Where to put abacavir? Where to put tenofovir? Each agent has been and can be used as the third drug or as part of the NRTI backbone.]

A1. First-line NNRTI therapy. We recommend first-line therapy with two NRTIs and an NNRTI - efavirenz once daily or nevirapine twice [or once?] daily.

B1. First-line NRTI therapy. We recommend first-line therapy with two NRTIs (from the above list) plus either abacavir or tenofovir.

B2. Second-line therapy. We recommend second-line therapy with a protease inhibitor and two new NRTIs. Nelfinavir can be used in most settings, though it may cause diarrhea. Other PI regimens require plentiful water supply and clean water (for indinavir) or accessible refrigeration systems (for ritonavir-containing regimens), we recommend the use of ritonavir-boosted indinavir, ritonavir-boosted lopinavir, or ritonavir-boosted saquinavir, each given twice [once?] daily.

B2. Second-line therapy. If a triple NRTI regimen is first (two NRTIs from list above plus either abacavir or tenofovir), it can be followed by either an NNRTI or a PI containing regimen.

C. Protease inhibitor regimen sequencing. If the PI regimen is first, it can be followed by a NNRTI regimen, and vice versa. If nelfinavir was used as the first PI, a second PI containing regimen may be effective following failure of the nelfinavir containing regimen.

Provide clearer recommendations on once-daily combinations. In the table, some drugs listed as once-daily do not yet exist (Zerit XR) or have yet to be validated (3TC, NVP, ???).

Something needs to be said about:

1. Preparing the patient for therapy
2. Ongoing adherence counseling before and during therapy
3. Clinical monitoring for side effects of therapy, and drug-drug interactions
4. Clinical management of drug side effects, including substitution of individual drugs or use of adjunctive therapy such as loperamide (Imodium)
5. Counseling the patient that therapy does not reduce the need to use barrier methods to prevent HIV transmission

Somewhere -- in the main document or an annex -- specific information needs to be made available on each drug - dosage, schedule, requirements for storage or for administration (e.g., with water, with food), drug interactions, side effects, toxicity management.

Section IV. When to Change Therapy

Here it will be difficult to maintain developed world standards except when CD4 counts and viral load tests are periodically available. What has been learned from the reduced use of these tests in settings such as Brazil or Uganda when people are given HAART but monitored less frequently, and presumably have more frequently broken through virologically, immunologically or clinically before switching therapies?

Switching for toxicity. In the US, switching for toxicity is at least as common as switching for virologic failure.

If switching for toxicity, identify the agent responsible and substitute an equivalent agent, e.g.:

• For hepatitis or severe rash on nevirapine, switch to efavirenz.
• For nephrolithiasis on indinavir, switch to another PI.
• For peripheral neuropathy on ddI/d4T, switch to AZT/3TC.
• For anemia on AZT, switch to d4T.
  Specific switch examples would need to be given for all drugs used.

Switching for failure (virologic, immunologic, or clinical)

1. Optimal setting. Viral load and CD4 counts every 3-6 months.
   • If VL, after having gone below 500 copies/mL, goes above 10,000 copies, assess adherence. If adherence is OK, switch entire regimen.
   • If CD4 fails to rise, or drops abruptly, consider switching.
2. CD4 testing available. CD4 tests every 3-6 months (or every 6-12 months).
   • Compare post-HAART CD4 to baseline. If it has risen, stay on regimen.
   • If post-HAART CD4 counts fail to rise, assess adherence. If adherence is OK, consider switching.
   • If post-HAART CD4 cells drop after rising, consider switching when they drop back to ??? what level above baseline or by what # or %?
   • Use of trimethoprim-sulfamethoxazole when CD4 counts drop below 200.
   • Develop plans to scale up viral load testing and availability.
3. Clinical monitoring. CD4 testing not available.
   • Immediately after starting HAART. If patient is sick, stay on therapy.
   • Immediately after starting HAART. If immune reactivation disease occurs, stay on therapy.
   • Later after starting HAART. If patient gets sick again, consider switching.
   • Later after starting HAART. If a previously asymptomatic patient becomes symptomatic or develops an AIDS defining condition, switch therapy.
   • Develop plans to scale up CD4 + viral load testing and availability.

Section V. Recommended Second Line Regimens

This section is dense, complex, and poorly written. Some of the writing here (e.g., first full paragraph, page 14) reads like an interior monologue of a resistance-obsessed virologist. Why not simply refer to the first- and second-line regimens mentioned above? How much resistance point mutation information do policy makers (or for that matter, physicians) need to switch entire regimens when the choices are all-new, generally non-cross resistant drugs.

A2 Switch from NNRTI based regimen to PI based regiment with two new NRTIs
B2 Switch from triple NRTI - PI or NRTI based regimen with two new NRTIs
C2 Switch from PI based regimen to NNRTI based regimen with two new NRTIs or
C3 Switch from nelfinavir-based regiment to another PI based, rtv-boosted regimen, with two new NRTIs if available

Section VI. Drug Resistance

Clarify, simplify.

Section VII. Special Populations

1. Pregnant Women or Those Seeking to Become Pregnant
   This section is missing in the draft I received.

2. Infants and Children
   This section is missing.

3. Coinfection
   Tuberculosis
   The tuberculosis section is clear, well written, and highlights the controversy over whether to initiate HAART when TB therapy is given or after it has been successful. HAART complicates TB treatment but may improve the success rate. Delaying HAART may increase TB related mortality. However, some HIV+ patients with active TB may, after their TB is cured, have an asymptomatic course for several years. Thus the management of TB coinfected patients may vary based on CD4 cell counts. However, little data are available to address which approach should be taken. I believe a study is being planned to compare TB therapy followed by HAART with TB and HAART currently. In any case, guidelines need to be developed now before the results of such studies are available.

   The authors recommend that if concurrent antiretroviral therapy is given with TB treatment, it should be a dual NRTI regimen with efavirenz, or a triple NRTI regimen. This seems to make sense given what we know about the adverse interactions of rifampin with protease inhibitors and with nevirapine.

   The authors also discuss the appealing idea of integrating delivery of treatment services for TB and HIV. While this has obvious appeal, it also raises the problem of exposing TB-negative HIV-positive individuals to TB on an ongoing basis while receiving treatment for their HIV. Practical steps will need to be taken to ensure that HIV-infected individuals do not contract tuberculosis while receiving HIV care.

Immune Reconstitution Syndrome
This is a clear, simple discussion.

Opportunistic Infection Prophylaxis
The authors appear to believe that "considerable controversy exists over the role of co-trimoxazole in the developing world." Why this controversy persists is a mystery I will never understand. A multitude of randomized studies in the developed and developing world show conclusively that this cheap agent can cut AIDS death and hospitalization rates in half, even in the absence of HAART. South African researchers discovered that many cases of Pneumocystis carinii pneumonia (PCP), once thought rare in Africa, had been mis-diagnosed as TB among AIDS patients, and that a retrospective chart survey of co-trimoxaxole use showed a significant reductions in mortality, confirming other randomized trials conducted in Côte d'Ivoire. Fears that appropriate use of co-trimoxazole would lead to an increase in drug resistant micro-organisms are as misguided a reason not to use it in HIV infection as fears of antiretroviral resistance are an excuse not to provide any HAART at all. Co-trimoxazole should be recommended for most patients with symptomatic HIV disease, and for all with CD4 counts below 200/mm³.

The authors should clarify which TB prophylaxis regimen they recommend. This paragraph should be moved up to the TB section, and the section here entitled "Co-Trimoxazole Prophylaxis."

Injecting Drug Users
This section is a bit wordy and confusing. It raises important issues and defends the equal rights of injecting drug users to access to effective care. Do we really need separate discussions here (in section iii) of specific HIV treatment dosage regimens for IDUs? I think this detracts from the overall consistency and simplicity needed in a document of this sort.

The discussion of the importance of appropriate analgesic use in all populations, including IDUs, is good.

Section VIII. Monitoring of Antiretroviral Therapy

The adherence discussion is good. Drug-drug interactions, toxicity monitoring and side effects management need to be discussed here, or somewhere.

Section IX. Conclusions

This is surprisingly well written, considering what has come before, and would even serve as part of a decent introduction.

Tables

Table 1. Recommendations for initiating antiretroviral therapy. It should say:

• AIDS
• Symptomatic HIV disease
• Asymptomatic disease with CD4 counts < 200/mm³

[Currently it does not say AIDS except in a footnote.]

Should the "Symptomatic HIV disease" footnote also include "diarrhea not due to other factors" and "oral hairy leukoplakia"?

Table 4, Antiretroviral Regimens for Initial Therapy -- Pros and Cons
Again, what is the point of these guidelines? They define a stance on these issues and lay out recommendations, rather than just saying "He said, she said." The table leads to the even more confusing and complicated following tables.

No simple recommendations are made anywhere in tables 5-8, unlike those I made above (in section III) or like the comparatively simple, easier to read Table 12, "Recommended Regimens for Initial Treatment of Established HIV Infection" in the US HHS Guidelines (HHS 2002, p. 43). Instead we get a list of:

• Twelve PI based regimens
• Eight NNRTI based regimens
• Four triple NNRTI regimens (all abacavir containing, no tenofovir)
• And a partridge in a pear tree! (Not really)

I don't know what I'd do with these regimens if I was a Minister of Health, an AIDS doctor, or a person living with HIV in a developing country and trying to make an informed decision. All those + and - signs in their inscrutably similar columns would simply make me throw up my hands and let the decision be made by the throw of the dice. Certainly it seems that by this point those who drafted these guidelines were no longer thinking clearly and rather than making firm recommendations, decided to make vague comparisons without enough specificity to be useful. These tables (5 through 8) contain more data on resistance point mutations -- information which will not be ready at hand in most places where these guidelines will be used -- while specifying nothing about the toxicities of the combinations being compared.

Finally with Table 9, "Recommended First-Line Antiretroviral Regimens," we get some specific recommendations. This table, or whatever replaces it, should go before tables 5-8 (if they are kept at all).

There really is no reason for AZT/3TC to be the only first-line NRTI combination in all three class-based regimens. d4T/3TC would certainly do as well. This leaves everyone with the necessity of taking d4T/ddI in their second regimen, perhaps leading to an excess incidence of neuropathy, pancreatitis and, possibly, lactic acidosis. Some thought should be given to having a bit less uniformity up front to preserve flexibility of options later. This would then affect Table 10, "Recommended Second-Line Regimens."

Finally, many of the drugs listed as "indication or potential use for once-daily administration" have not yet been developed (d4T-XR), or have not yet to my knowledge been indicated for once-daily use (3TC, ABC, NVP, and all three of the ritonavir boosted regimens).

The authors of the WHO Guidelines have more work ahead of them. Their tasks now are to clarify their objectives, simplify their text, make firmer recommendations on whom to treat and how, and to demand that the WHO, UNAIDS, the Global Fund for AIDS, TB and Malaria, and all the other players in this dystopian bureaucratic morass mobilize more effectively for a massive scale-up of HAART treatment. It's time for WHO and the others to make a clear commitment to have treatment programs underway for five to ten million HIV-infected people worldwide within five years.