The NCI Randomized Trial of High Dose IL-2 vs. Low Dose IV IL-2 vs. Low Dose Subcutaneous IL-2 for Metastatic Renal Cell Cancer

One of the few important randomized trials involving high dose IL-2 was conducted by Dr. James Yang and his colleagues at the US National Cancer Institute (NCI) starting way back in 1991. The final results were not reported until late 2003 [Yang 2003]. This trial involved 400 patients. This should give you an idea how much time and effort it takes to answer questions about treatment differences with treatments like IL-2 for which the benefit is long term survival for a minority of patients.

Dr. Yang and colleagues have been very open about the progress of this trial over the years. They presented data at several meetings and actually published an interim analysis in 1994 [Yang 1994], and another in 1997[Yang 1997], long before they published the final results in 2003 [Yang 2003]. Very interestingly, most of this took place even while the trial was still open to new patients, and in fact I've been following and reporting on this trial for years. Dr. Yang's openness gave patients intrepid enough to delve into the literature (or read my page!) a unique opportunity to get an idea of the way the results were trending as they considered enrolling. While this situation isn't common, it's always worth asking whether any preliminary results are available for trials you're considering.

Trial Design

This randomized trial had three arms:

  • Standard High-Dose IV IL-2 (720,000 IU/kg/Dose).
  • Inpatient IV IL-2 at 10% of the standard dose (72,000 IU/kg/Dose) but the same schedule as standard high dose.
  • Outpatient Subcutaneous IL-2 at 250,000IU/kg/day for 1 week, thereafter 125,000 IU/kg/day for 5 weeks.


  • Additional courses of treatment were given to responding or stable patients. For all arms there was a rest period after treatment long enough to make a total of 8 weeks before starting another cycle of treatment. For the IV arms, this is about 5 weeks of rest, and for the outpatient arm, two weeks.


  • All patients were in good enough shape to take high dose IL-2. This was necessary because patients could be randomized to that treatment. Importantly, they specifically limited the trial to patients with clear cell RCC (This is by far the most common sub-type, but see my article on sub-types for details). This means the results only apply to patients with clear cell RCC.

Initially the purpose of the trial was only to compare high and low dose inpatient IL-2, and so there were only two arms. The treatment in the low dose arm is the same as standard high dose IL-2 except at only 10% of the dose. The investigators chose this dramatically lower dose because it is the highest dose which avoids severe side effects. If lower dose IV IL-2 is at least as effective as high dose IV IL-2, then it would be preferred since, as the trial data clearly shows, the side effects are substantially less severe for the low dose regimen. Low dose IV bolus dose IL-2 has not been a frequently used treatment, so a finding that high dose is better would not have a big practical impact, although it might be interesting scientifically.

Later, because outpatient IL-2 therapy had come into widespread general use and can also be effective, the investigators added a third arm for subcutaneous outpatient therapy. The side effects of outpatient IL-2 are also less severe than for the high dose regimen, but, as Dr. Yang points out, the outpatient treatment goes on longer and takes up a higher proportion of the time the patient is on treatment in comparison to the IV regimens, where the treatments are short and the time between courses of treatment is much longer. So it is unclear from the patient perspective whether lesser side effects for longer, or more severe side effects for a shorter time would be a better choice if they were both equally effective. I think it would be personal preference. Still, it is important to know if this commonly used outpatient treatment is as good or better than the standard regimen, considering that far more patients get outpatient IL-2 therapy than high dose, and that high dose IL-2 is more expensive and difficult to give.

In comparisons involving the outpatient arm, only patients who were randomized after this arm was added are included, but in comparisons of the original two arms, all patients randomized to either of these two treatments, including those who entered after the subcutaneous arm was started, are included. This guards against bias from any unrecognized change in the characteristics of the patients or the treatment over time, since only concurrently randomized patients are compared. The disadvantages are that the three arm comparison involves fewer patients in each arm than the two arm comparison (and thus less reliable results), and also that follow-up time is shorter for the three arm comparison.

Response Data

Below are all of the response data for both the two and three arm comparisons. These data pretty clearly favor high dose.

Response Data: 2-Arm Trial
 High Dose IV ArmLow Dose IV
Number of Patients156150
Number of Complete Responses (%)11 (7%)6
Number of Partial Responses (%)22 (14%)13
Total Complete and Partial Responses (%)33 (21%)19
Duration of Complete Responses (months)130+a, 121+, 115+, 114+, 103+,
100+, 90+, 52+, 45, 23, 19
128+, 113+, 40+, 20, 19, 3
Duration of Partial Responses (months)37, 28, 24, 23, 19, 17,
17, 16, 15, 14, 14+, 13, 10, 9, 8+, 8, 8, 7, 6, 4, 4
24, 23, 22, 21+, 15,
13+, 11, 11, 8+, 7, 7, 4, 4

(a) Note that the plus sign means the response is still ongoing.
Conversely, the lack of a plus sign means the patient relapsed after the given time.

Response Data: 3-Arm Trial
 High Dose IV ArmLow Dose IV
Subcutaneous Arm
Number of Patients969394
Number of Complete Responses (%)6 (6%)1
2 (2%)
Number of Partial Responses (%)14 (15%)9
7 (7%)
Total Complete and Partial Responses
20 (21%)10 (11%)9 (10%)
Duration of Complete Responses (months)103+, 100+, 90+, 52+,
45, 23
40+78+, 13
Duration of Partial Responses (months)37, 28, 24, 23, 19, 17, 17,
14, 14+, 10, 9, 8+, 8, 4
24, 21+, 15, 13+, 11, 8+, 7, 4, 428, 28, 17, 15, 9, 8, 2

Survival Data

[Yang 2003] Figure 1a, Survival for Low Dose IV vs High Dose IV


  • First, this confirms the important characteristic of IL-2 treatment. There were long term complete responders in each arm of the trial (though very few for the low dose arms). Some of these responses are ongoing after in excess of ten years. These patients are probably cured.
  • Second, as you can see, the response data strongly favors the high dose arm, especially if you look at the responses which really lasted. In this study, it doesn’t look like any of the partial responses were truly long term. A few are still ongoing, but only with relatively short follow-ups, and the pattern is that they likely won’t last. Despite this, some of the partial responders got a remission two or three years long. That is a significant benefit. The differences in response rates were borderline statistically significant. Overall, the pattern is enough to conclude the differences are likely not due to chance alone, and that the high dose treatment produces more responses, and more durable complete responses. In short if you are going for a cure, you’re more likely to get it from the high dose treatment.
  • Third, despite the difference in response, differences in long term survival, as shown by the right ends of the survival curves were small (and surely not statistically significant) for the two arm comparison, and essentially non-existent for the three arm comparison. If response to IL-2 leads to long term survival, and high dose IL-2 leads more long term responses, then why didn’t high dose IL-2 lead to better long term survival? It turns out that this can be explained by the existence of long term survivors who did not have a durable response. It’s important to know that responding to IL-2 isn’t the only way people became long-term survivors.

Survival Versus Response: An Extended Analysis

It’s well known that temporary responses (especially to chemotherapy) may have very little effect on survival, so oncologists are usually skeptical of differences in response rate without differences in survival. But here, the complete responses included many which were long term. By definition these affect long term survival. There were lots more durable complete responses for high dose than for either low dose treatment. Therefore, if the survival curves do not show a clear difference there must be other long term survivals which were more common in the low dose arms. It turns out this is the case, and not only that you can actually see it in the data!

If you look at the survival curves closely, you can see vertical tick marks which indicate where the follow-up of individual survivors ended with the patient still alive (see my CancerGuide article on Kaplan-Meier Estimation for more details of what the tick marks mean). You can get an idea of the number of long term survivors by counting tick marks towards the end of the curve. You can then compare this to the number of durable responses for each arm and figure out how many patients were long term survivors even though they didn’t have a durable response. I defined long term as 5 years or more and then did the math. Note that a little bit of what I report here is only deducible from data in the paper about surgery for relapse after complete response. Also note that the curves were not plotted out to the length of the longest survivor, and therefore, the end of the curves may hold a small number additional number of long term survivors. Finally, note that there was a small number of patients with ongoing complete responses but follow-up of less than 5 years, for example, the patient in the low dose IV arm who is in complete ongoing response at 40 months. These patients may well become long term survivors, but I am not counting them as such here.

Analysis of Long Term Survivors: 2-Arm Trial
 High Dose IV ArmLow Dose IV Arm
Long Term Survivors (from counting tick marks)128
Long Term Responses (from response data)8a2
Long Term Survivors Without Long Term Response56

(a) This includes a complete responder who relapsed, but is a long term survivor after being rendered disease free by surgery. Another complete responder who relapsed is also currently disease free by surgery, but without enough follow-up to be included as long-term here.

Although these numbers are tiny, the survival due to response favors the high dose arm 8 to 2. If you look at the ends of the survival curves for the two arm trial, you can see that the high dose arm is only a little better in the end. From this table, you can see that the difference in survival due to response is diluted by a significant number of long term survivals without response (about equal in each arm). It’s also very interesting that the total number of long term survivals without long term response (11) is actually about equal to the number due to long term response (10).

Analysis of Long Term Survivors: 3-Arm Trial
 High Dose IV ArmLow Dose IV
Subcutaneous Arm
Long Term Survivors (from counting tick marks)425
Long Term Responses (from response data)301
Long Term Survivors Without Long Term Response124

Here the numbers are even smaller, but the total number of long term survivors without long-term response (7) is actually more than those with long-term response (4).

For both comparisons, the absolute number of long term survivors is so low that random distribution of the long-term survivals not due to response can easily dilute or overwhelm any effect of the treatment which produces more survivals in one arm. This again points up the difficulty of testing treatments which produce a small number of cures but don’t benefit most patients.

I am always interested in long term survival, so I was intrigued by the existence of these mystery survivors. I wrote to Dr. Yang who told me that these other long term survivals are due to a number of different causes, including patients who benefited from other clinical trials after IL-2 failed, patients who had all tumor removed surgically after getting a partial response from IL-2 and then relapsing, as well as several other causes. (James Yang, MD Personal Communication, October 2003).

This variety of paths to long-term survival gives no cause to suspect that any of the treatments would be more likely to produce this kind of survivals than another. I believe then that random distribution of these “other” survivals totally obscured any benefit of the high dose arm in the three arm survival curve, and diluted the benefit in the two arm survival curve (where the “other” survivals appear to be about equal between the arms).

There is another and equally important conclusion to be drawn here. It is important for you to know that even if IL-2 does not work for you the game is not over. There are a variety of paths to long term survival including surgery and enrollment in carefully selected high quality clinical trials. This represents additional hope concealed within these statistics.

Given the variety of these other paths to long term survival, their existence, while offering hope, does not change the fact that IL-2 was surely still the single most likely way to become a long term survivor in this study, and high dose IL-2 appears more likely to yield that result than low dose. Also, while we know the long term survivors who had a complete response to IL-2 have no detectable cancer (and are probably cured), we don’t know the status of the other long-term survivors.

Special Note on the Role of Surgery in this Trial

I view integration of surgery with immunotherapy as essential to maximizing the benefits with immunotherapy, which also means it can have important effects on trial results.

Almost all of the partial responders relapsed and none have been proven long term durable. Yet other studies have shown that when all disease can be resected after a partial response, long term relapse-free survival is common See my article on combining surgery with immunotherapy for details.

Reading this paper, I could not find anything to suggest they attempted to remove residual disease after a partial response, except in the case where the only remaining disease was the primary (kidney) tumor. I confirmed this with Dr. Yang who told me that with this one exception, their policy is to consider surgery (or other treatment) only after the disease has started to progress (James Yang MD, Personal Communication, October 2003). It’s impossible to say how many patients might’ve been eligible for surgery who didn’t get it, or to know how often waiting for relapse makes surgery that would have been possible, either impossible or ineffective, but I think it is possible that some additional patients might have been able to achieve long-term survival through surgery combined with the immunotherapy. Based on my reading of other studies, I think it’s possible rigorously combining surgery with immunotherapy would have significantly improved long term results.


The paper [Yang 2003] concludes:

… in the absence of definitive differences in survival, low dose
IL-2 remains a viable therapeutic option for patients with significant medical
comorbidities, or for physicians without experience in giving high-dose IL-2.
When patients and their treating physicians are able to pursue a high dose IL-2
regimen with a risk of irreversible toxicity or death that is less than 1%,
then administering IL-2 at high doses to patients with metastatic clear-cell
renal cancer should be the therapy of choice.

I agree with this but would go a little further. As an active patient you have the option of seeking out high dose IL-2, if (as is likely) your doctor is not offering it. It’s worth restating that it is best to get this treatment from a real expert. An expert will not only know when to stop, but also when to push on to deliver the most treatment.


Yang JC, Sherry RM, Steinberg SM, Topalian SL, Schwartzentruber DJ, Hwu P, Seipp CA, Rogers-Freezer L, Morton KE, White DE, Liewehr DJ, Merino MJ, Rosenberg SA.
Randomized study of high-dose and low-dose interleukin-2 in patients with metastatic renal cancer.
J Clin Oncol. 2003 Aug 15;21(16):3127-32. [PubMed Abstract (will open in new window)]

Comment: This is the final report.

Yang JC, Rosenberg SA.
An ongoing prospective randomized comparison of interleukin-2 regimens for the treatment of metastatic renal cell cancer.
Cancer J Sci Am. 1997 Dec;3 Suppl 1:S79-84. [PubMed Abstract (will open in new window)]

Comment: This is the final report.

Yang JC, Topalian SL, Parkinson D, Schwartzentruber DJ, Weber JS, Ettinghausen SE, White DE, Steinberg SM, Cole DJ, Kim HI, et al.
Randomized comparison of high-dose and low-dose intravenous interleukin-2 for the therapy of metastatic renal cell carcinoma: an interim report.
J Clin Oncol. 1994 Aug;12(8):1572-6. [PubMed Abstract (will open in new window)]

This CancerGuide Page By Steve Dunn. © Steve Dunn
Page Created: February 26, 2003, Last Updated: February 3, 2004