To tell if it's time for a new cocktail, take the tests, listen to your HAART, and learn what undetectable really means

Is your drug combo working? Of course, this question assumes that the tradeoffs between the combo's antiviral punch and the quality-of-life concessions are acceptable. That said, the simplest measures of whether your regimen is running at top speed are those old standbys, CD4 count and viral load. But never just check a single test's snapshot -- track the trend over time.

CD4 cell counts are far more reliable than viral load. A main marker of immune health, this number flashes "danger" at fewer than 200, when the risk of opportunistic infections and cancers rises. While there are exceptions -- people with PCP at 400 CD4s, say, or in top form with zero -- nearly 20 years bears out this rough parameter.

In contrast, experts are just beginning to appreciate the caprices of HIV viral load. For some HIVers, a mere 10,000 to 20,000 viral copies (per milliliter of blood) appears sufficient to cause immune deterioration, while for others, a viral load teetering at 100,000 or more inexplicably co-exists with normal CD4 counts -- for years! Viral-load hawks argue that changes offer an early and accurate forecast of what will happen to CD4 cell counts, but this theory remains sketchy.

For HIVers just starting a standard three- or four-drug combo, the goal is to suppress viral load below the limit of detection (400 copies/ml) within four to 12 weeks. But if your test is pricey and ultra-sensitive, measuring down to 20 copies, reaching undetectable could take twice as long. So if you're starting with a viral load of, say, 500,000, give it six months or so before fretting.

Does failure to reach undetectable by deadline mean the jig is up? It ain't always so. If you're taking your first cocktail as directed and you aren't undetectable by week 16, you need to think about "treatment intensification" (see below). Or you could trade in the entire combo and start from scratch.

That's the easy part. For the therapy-experienced HIVers, the definition of "failure" depends on who's doing the defining. At the dawn of protease hype -- and viral-resistance hysteria -- any post-detectable HIV breakthrough above 10,000 or so was deemed proof positive that the wily virus was headed toward uncontrollable resistance. This exuded drama on par with ER: "Dr. Benton! We have a viral rebound in Room 3! Nurse Hathaway, bring the salvage-therapy cart. Stat!" Mellower these days, docs appreciate that treatment failure comes in many sizes.

Your combo's days may be numbered if your viral load has (a) become repeatedly detectable after being undetectable, (b) returned to baseline (the level when you started treatment) or (c) broken through a significant (20,000 to 50,000) threshold of replication. Still, many specialists are cautious about a quick switch to a new combo, both because breakthrough replication often plateaus at a manageable level and because of the limited number of options.

So, if you've previously been undetectable and -- gasp! -- suddenly get a "detectable" result, what to do?

1. Change nothing. But monitor closely and wait. This is particularly true when you have so-called viral blips, minor bursts of viremia, in the double or triple digits, one time only. A repeatedly detectable viral load is probably not a blip but a sign of resistance. Also check for absorption or interaction among HIV meds and other drugs.

2. Intensify. Treatment intensification -- adding a drug to ratchet up a combo's antiviral punch -- is quite the buzzword. Low doses of ritonavir have been shown to increase the way other PIs are broken down in the body. This not only helps push a rebounding viral load back down but greatly simplifies a regimen. For example, low doses of ritonavir increase indinavir levels almost fivefold, while decreasing dosage to twice a day.

3. Ferret out the offender. Typically, it's only one (maybe two) drug that stops working when viral load rears its ugly head at your combo. Drug-resistance tests can help determine which med has lost its luster. The key is to act fast. If only the PI is fingered as suspect, swapping it for either a PI with different mutations or an NNRTI is usually best.

4. Change the regimen. While not the most resourceful option, a total makeover in the face of rebounding virus is the safest bet. This is relatively simple for folks failing a first regimen -- just try other drugs in the same classes. After that, though, it might be necessary to recycle old drugs and work around mutations that confer cross-resistance.

Despite all the promise, the practical utility of drug-resistance tests -- genotypic or phenotypic -- remains questionable. Apart from helping select a first regimen for people infected with an already-resistant virus, the tests so far seem most useful for HIVers still on their first combo or naïve to one class of drugs. Obviously, this narrows the field a bit.

While early clinical trials tried to prove that these pricey tests were superior to clinical history and the judgment of in-the-know docs, the latest man-vs.-machine studies suggest that the mortal brain is as effective as tech brawn. Also, resistance test results are often misinterpreted because they serve as negative indicators: They rule drug(s) out, not in. The fact that a mutation doesn't show up on your genotype readout offers no info about whether you will respond favorably to that drug. Phenotypic tests, while more useful in this case, are still far from perfect.

Two high-tech tests may eventually revolutionize treatment decisions by giving a more precise measurement of a drug's effectiveness in your body. Drug-level monitoring ("therapeutic drug monitoring") counts the amount of drug in the blood, but the test isn't standardized or readily available, and each class of drugs has drawbacks that make the results difficult to apply. Drug-cell monitoring, more promising because it measures the concentration of drugs within the cells, is even further away. So don't hold your breath -- take a deep one, forget the tests, and treat yourself to a real cocktail for a change.