Saturday, August 14, 2004

Why Viral Loads are Useful, Relevant and Important.

A commonly spread myth is that the HIV viral load tests are meaningless, non-specific and arbitrary. This myth is spread based on several lines of logic that are either flawed, misleading, or just downright lies.

Firstly a bit of background as to what a viral load test is. Simply put, a viral load is a measure of the amount of genetic material in the bloodstream, as judged by the relative amount of certain RNA molecules. In the context of HIV, viral load is used as a confirmatory test of infection, but more usually as a judge of how rapidly the disease is progressing. The higher the viral load, the faster the infection is likely to progress to AIDS, and ultimately death. The aim of anti-retroviral therapy is to reduce the viral load to below detection (less than 50 copies per ml with many modern tests).

The Dissidents do not like the test for several reasons, as discussed below.

1: “The VL test doesn’t look for actual virus, it only looks for RNA”.

This is true, but a molecular biologist would simply say “So what?” There is nothing wrong with this, no matter how many dissidents say there is or how loudly they say it. Ultimately unless you physically purify HIV from peripheral blood (see the email from Gelderblom as to why that isn’t likely with HIV) and look at it under EM, you are limited to using secondary detection measures. These will include looking for viral proteins or genetic material, or virus culture. It is a standard procedure for a great many other viruses, and increasingly for some bacteria. This argument is a bit like saying “There’s no proof I was at the scene of the crime, they’re just my fingerprints”.

2: “The VL is amplified many times then a number is back-calculated”

This isn’t really true. It is mind-numbingly obvious though that amplification HAS to be done. In order to visualize the RNA on a standard lab gel you would need to bleed 20 AIDS patients completely to get enough virus to barely see the RNA from all the virus in their bloodstreams (6.5ng). About 200 patients to make it believable at 65ng (a nanogram is a billionth of a gram) – there simply isn’t enough material. The “back calculation argument” though is very misleading. Due to the variability in this kind of amplification (called PCR – polymerase chain reaction) you must have some kind of controls. Typically these are internal controls, whereby samples of known copy-number are run at the same time in the same conditions. Then a graph can be drawn of all the “standards” and the experimental sample is then fitted to the graph. It is a common Dissident tactic to either purposefully (or through ignorance) leave out the fact that the proper controls are always done. It would be impossible to directly count the RNA strands after the amplification process has been done, which is what dissidents imply needs to be (and is) done. Instead, standard molecular techniques are used (where for example the concentration of RNA is shown by how absorbent the sample is for a certain color of light, compared to a control sample).

3. “You only look for a small part of the virus, not the entire genome”

This is true, but again the answer has to be “So what?” The big concern is that if you look at too small a sequence you might by accident discover an RNA molecule that looked similar to HIV in that small sequence. The HIV genome is about 10,000 bases long, and the viral load detects bands that are only a few hundred bases long. In order to get enough RNA to detect (actually DNA, since they copy the original RNA into DNA first because it’s far, far easier to copy and work with) many rounds of amplification have to be performed, around 20. Copying the entire 10,000 base genome would take around 20 minutes to perform per cycle, this means that with the rest of the protocol the entire process would take about 10 hours to complete! While not impossible, by limiting the process to only a few hundred bases it can be complete in an afternoon. However, by arguing that longer stretches need to be used the Dissidents only highlight their ignorance of a vital step in PCR – no matter how long your sequence is, the actual copying process is started by short stretches of DNA called oligos, which are typically 20 or so bases in length,

As such, it makes no difference whether you copy 100 or 10,000 bases, you’re going to have the same degree of accuracy, since the limiting step is the uniqueness of the oligos!

As for the specificity of the oligos, you can get a measure of how specific they are by how long they are. The chances of a single base matching are 1 in 4, since there are 4 different bases in RNA and DNA (A, C, G and T/U). The odds of two consecutive bases matching are 1 in 16. By the time you get to 20 bases you have a 1 in 1,099,511,627,776 chance of matching, or over 350 times more DNA than is in the entire human genome! Obviously certain sequences will be “conserved”, meaning their sequence is more likely to occur than pure chance would suggest. However, despite the existence of retrovirus-like sequences in the human genome (called endogenous retroviruses, HERVs), this is still very unlikely if you pick the right place to look. If you couple that with the fact that you are expecting a product of a certain size (say, 300 bases) you can be fairly sure that no other sequence of DNA, no matter how long, is going to bind to the oligos you use to produce a product of that size.

It is true that the virus RNA is only 1% of the RNA produced by an infected cell. By virus standards though, that it pretty high! That means you really only have to secure uniqueness to 99% of the RNA (if the ratio was the same as the length of genomes, you would need to show uniqueness to a 3 million-fold excess! It is exceedingly unlikely that any RNA detected would be from non-HIV sequences, and any that are would be very low levels. Normally RNA is not found outside the cell in appreciable numbers, whereas of course virus particles are definitely outside the cell! This further increases the likelihood of distinguishing HIV from any other sequence.

4. “The RNA doesn’t equate to numbers of infectious virus”

This is true, and contrary to what the Dissidents say is not assumed to be such by the scientists (an example of them putting words into our mouths). It is well recognized that 1 infectious virus equates to about 60,000 viral genomes per ml. It doesn’t matter, since we’re not trying to say that the virus in the bloodstream is causing AIDS. What?! I hear you cry, surely that’s what they say? No, that’s only what the Dissidents say the scientists say, another example of putting words into our mouths. Viral load is used as a marker because of very good reasons (detailed below) which do NOT include the assumption that the viral load is causing AIDS. In fact, it is very well recognized that the virus in the bloodstream is only a tiny portion of the virus in the body. The vast majority is to be found in lymph nodes, which is where AIDS really happens. Obviously a blood test is far easier than taking a lymph node biopsy, hence the reason why VL measurements are used instead of LN biopsies.

But how much more virus is there in lymph nodes? Far, far more than is in the bloodstream. Maybe 1 in 40 CD4 T cells in the bloodstream contain HIV in end-stage AIDS, far fewer in earlier stages of infection. In the lymph nodes maybe 1 in 4 T cells is infected! Since only 10% of T cells are in the bloodstream, it’s clear that only a small part of the virus replication goes on in the bloodstream. There are plenty of reports suggesting that LN biopsies are used as measurements of disease progression, at least there were in the days before viral load measurements!

On the other hand, it’s obvious that the relative amount of infectious RNA must equate to the viral load. Even with 1 in 60,000 being infectious, if a viral load increases 10-fold then the number of infectious virus must increase 10-fold. A small proportion does NOT equate to NO infectious virus! Having a low infectivity isn’t unusual for viruses (herpes is about 1 in 100), but HIV is admittedly worse than many. However, implying that every copy of the genome should be infectious isn’t right. This point also destroys another Dissident argument: that HIV cannot be found in infected people. Clearly if a ratio of 1 in 60,000 can be measured, is MUST be true that HIV can be cultured from patients! In fact, HIV can be cultured from all AIDS patients, and not from people without HIV antibodies.

In any case, it is known that non-infectious HIV can cause disease! The proteins contained in and on the virus have been shown to cause immunosuppression and cell dysfunction, as detailed in my refutation of Duesberg’s work.

5. “The Viral load is based on detecting RNA that hasn’t even been shown to be that of HIV”

This is blatantly false. Contrary to whatever people like the Perth Group say, HIV has been purified enough to know that the RNA molecule labeled as “HIV” is in fact that of a virus particle. The RNA encodes proteins that co-purify with it (and not cellular proteins), which to any virologist is practically 100% proof that it belongs to a virus. The fact that the Perth Group are not virologists is the only reason why they say the things they do. They quite simply have no idea what they are talking about.

6. “Anti-HIV medications reduce viral load by reducing levels of cellular RNA”

This is ludicrous. Viral loads can be reduced 100,000 fold by antiviral therapy. If cellular functions were reduced by that level you would die in very short order! Any “escapes” from therapy are correlated with mutations in the HIV proteins and RNA. This should not be happening if the effects were non-specific. Contrary to what the Dissidents say, the meds are very specific indeed. 100 to 10,000 more drug is needed to affect cells compared to virus, and that’s with the most toxic medications. The protease inhibitors can not affect cellular RNA or DNA production at all!

7. “False positive results are common”

There are several documented false-positive viral load cases. Some have led to mis-diagnoses. However, they are not as common as is suggested. Viral load is often mentioned in the same breath as p24 antigen testing, which is very clearly a bad test for diagnosis or monitoring of HIV infection. Even with the fact that viral load is a far better test, it is not used for routine diagnosis of infection. It is usually used for judgment of treatment initiation or change, as discussed below. How do these false positive results occur? The simplest explanation is that of accidentally binding to a close match of DNA in the reaction, or contamination in the lab. Due to the sensitivity of the test, detecting contamination is entirely possible. False-positives are usually low levels, around 1000 counts.

8. “HIV DNA detection is very non-specific, why should RNA be any different?”

HIV DNA detection can be non-specific, depending on the oligos used. Some cross-reaction with the endogenous viruses can occur. However, they do not produce RNA or virus particles, since they are usually inactive, so by removing the DNA and using only extra-cellular virus particles you reduce the likelihood of detecting these endogenous particles. I have seen it written that the sequences looked for: Gag and Env are “specific to HIV”. No-one with any experience with virology at all would say that: Gag and Env are simply names for genes that encode the structural proteins of a retrovirus. Every retrovirus in existence has its own version of Gag and Env: it is not surprising that depending on how you look you can confuse HIV Gag and HERV-K Gag, for example.

9. “Viral load improvements with therapy are only due to placebo effect”

Placebo means “it pleases me”. It is a psychological phenomenon. It cannot explain molecular measurements like viral load, unless they were measuring some kind of stress hormone. It also doesn’t explain why people with naturally low viral loads have strong anti-HIV immune responses, and why rising viral loads match with mutations that either escape the drugs or the immune response.

10. Viral load levels predict disease progression. This is the biggest anti-Dissident point of them all. In many studies, it is clear that the best predictor of how soon someone will progress to AIDS is how high their viral load is. This fact makes the level of viral load very important, even if it weren’t measuring infectious virus, or even HIV at all! This is undeniable. A viral load of over 30,000 equates to a 50% risk of progression to AIDS in 2 years, independent of CD4 count. Lowering viral load does correlate with reduced mortality and opportunistic infections. Lack of control of viral RNA correlates with worse outcomes. What more is there to say…?


Blogger Unknown said...

Hi Bennett,

According to 1 above, people “are limited to using secondary detection measures” like “viral proteins or genetic material, or virus culture” to find HIV because it’s not likely to “physically purify HIV from peripheral blood.” If what you say is true, then how were these secondary detection measures determined to belong to HIV in the first place?

In #2 above you write about “samples of known copy-number” being used as a control in determining viral load. How are these samples of known copy-number counted?



12:31 PM  
Blogger Bennett said...

By most standards these days, an organism is defined by its genetic material. Bacteria and viruses are put into species and strains based not on what they physically look like, but on what their genetic material looks like at the molecular level (i.e. genetic sequencing). The consistent thing about this is that you can decode the genetic sequence into protein sequence. When you can co-purify physical evidence (EM micrographs, enzyme activity) proteins (by ELISA or Western Blot) and genetic material (viral genomes) that are internally consistent then you can safely say that they are all from the same organism. Whether you purify that organism from the host or from a tissue culture is irrelevant - you've still demonstrated internal consistency between those biomarkers.

To put it concisely, you will have shown that a physical structure exists that is made up of proteins that are themselves encoded by a genetic structure that is held within that structure. That is nothing less than a decent description of a virus.

Then, when you go ahead and use the viral-load or Western Blot to detect HIV in the blood you know that you are detecting HIV because your prior work has demonstrated this secondary measure as being uniquely associated with HIV and as being consistent with coming from HIV.

Contrary to dissident opinions and statements to the contrary, HIV's genetic material and proteins are not found in the human genome, and as such it cannot possibly be derived from the human genome. When they complain about proving whether or not these proteins come from HIV they forget that if the proteins aren't from humans, they must be from somewhere else...

The samples of known copy number are counted by the simply virtue of being mass-manufactured. There are very easy methods for determining amounts of nucleic acid, for example serial dilution, UV absorption etc. Either they directly tell you the number of copies present or you can work it out knowing how large each individual molecule is. The companies that produce viral-load assays will provide control samples of DNA/RNA (whether HIV-related sequence or not) to run alongside the sample specimen. The final result is a comparison of the sample to the controls.

This article describes one such approach to developing a standard curve for controlling for PCR efficiency issues. As you can see, AIDS dissidents do not have the lock on seeing the issues on whether or not a particular assay or experiment might have problems. However, they are arrogant enough to assume that they are the only people to see the errors! They are also somehow unable to accept that someone else might have, somehow, bypassed the problem.

Note that the "curve" is in fact a straight line because of the logarithmic graph. This in fact makes the standard-curve more accurate when trying to calculate a sample's value within that curve.

You can also appreciate that the "copy number" is, at best, a match between the sample result and the expected result from the standard curve. It isn't necessarily stating that there are exactly this number of copies present in the sample: merely it says that enough copies of sample sequence are present to correspond to a particular number of copies of control sequence. What it does mean though is that if your result is doubled, your original starting value must be doubled.

This is probably why the correlation between viable HIV virus and viral load number is such that it's far from 1:1 (closer to 1:10,000 or more). But for any situation, you can safely say that a viral load of 30,000 is twice as much virus as a viral load of 15,000, even though the assay isn't strictly telling you that there are definitely 30,000 copies of the HIV genome present.

Of course the assay developers try to make them as accurate as possible, so it's going to be close, but understanding and accepting that limitation is crucial to disregarding the denialists when they complain that viral load doesn't directly measure live infectious virus. My answer is "of course not, and so what?"

The real proof is in papers like Mellors and Rodriguez where they show that, on average, people with higher viral load assay results are more likely to progress faster to AIDS and die and that people with medically-treated undetectable viral loads live a longer time. Whether you're measuring live infectious HIV or moldy cheese, that makes viral load measurements pretty important.

4:54 PM  
Blogger Unknown said...

Thank you for your time. I never earned a PhD in the microbiology of HIV so forgive me if I fail to understand some things.

So far, I understand that no one physically purified HIV from peripheral blood. Someone cultured it in a lab.

I assume people used these cultures to determine the secondary detection measures for HIV. I wonder how.

The chapter “Concentration and purification of viruses” in the book, Virus Culture says, “… concentrated and purified preparations will be required for investigations such as… genetic analysis where the genome sequence is not known.”

Once upon a time no one knew the genome sequence for HIV. Who first determined this sequence and how?

I assume they used a concentrated and purified preparation of HIV. I also assume they took an EM picture of this preparation. If the people who first determined the genetic sequence of HIV failed to use a concentrated and purified preparation in their investigation, then please tell me what methods they used to determine HIV’s genome sequence.

Also, regarding my original question about samples of known copy number, I wonder if anyone ever used EM pictures of concentrated and purified preparations to count HIV. If so, who?

From the book Basic Virology, “The ability to count viruses ultimately depends on the ability to see them, and this requires special techniques that were not available until just prior to World War II. Notable among these is the electron microscope (EM). Since virus particles can be purified and visualized, they can be counted. All counts require visualization. Once the number of virus particles in standard solution is known, this information can be correlated with other readily measureable properties [secondary detection measures?] of the virion.”

If the first people who counted HIV particles failed to take EM pictures of a concentrated and purified preparation of HIV in their investigation, then please tell me what methods they used to count these particles.

So, using the books referenced above, I guess one needs a purified and concentrated preparation of HIV, including EM pictures of these preparations, as a starting point to determine secondary detection measures, like genome sequence.

But, if people failed to use concentrated and purified preparations of HIV and hence failed to take EM pictures of these preparations in their determination of these secondary detection measures, then how were these secondary detection measures determined to belong to HIV?

Thanks again!



5:39 PM  
Blogger Bennett said...


You've asked a lot of important questions. So important that in order to fully explain them I'm going to have to write a full post to include a few pictures :o) Bear with me and I'll try to get it done sometime over the weekend or early next week. Suffice to say that all of the steps you mentioned have been done, although some of your assumptions about how and when these steps were done aren't right.

I'll try to pull it all together - watch the main page for the new post...



11:00 AM  

Post a Comment

<< Home