A disease that afflicts only 350 children per year in the US (in the high-risk form) does not make headlines very often. But after the September 30, 2010 publication of the New England Journal of Medicine article revealing the results of the phase III chimeric antibody trial (ch14.18 given with two cytokines GM-CSF and IL2), neuroblastoma was all over the news including prime time national news. Over 200 news stories appeared within the next 2 days and over 3000 blogs reported on the story. Click on image below for a nice example of one of the medical blogs:
The news was actually first released March 19, 2009 after an early review of the study. The study was amended so that the randomization was stopped and all eligible children could receive the antibody.
This is quite a dramatic story on many levels.
An absolute must read is an excellent article giving more background on the story in the NCI Cancer Bulletin. The article details the incredible perseverance required of Dr Alice Yu, Dr Paul Sondel, Dr Malcolm Smith, and the entire COG team to bring this antibody to children with neuroblastoma. The research on this antibody began in 1985, and yet it took 25 years to get solid proof that the antibody improves survival. Why did it take so long?
Antibody and drug development are not the same
Antibodies are first isolated from mice that are “challenged” with a tumor and produce antibodies against that tumor. The production is shown in the illustration below from a wikipedia article which describes the process:
This particular antibody targets GD2 which is a glycolipid (sugar-fat) antigen on the surface of NB cells. This antigen is also present on other cancers, including melanoma. GD2 is also expressed on some normal nerve cells, which is why the treatment causes pain. “First generation” antibodies are entirely mouse products (termed “murine”) and is why a normal immune system reacts quickly to produce anti-mouse human antibodies (HAMA) which effectively neutralize the action of the mouse antibody. Examples of first generation anti-GD2 antibodies are Memorial Sloan-Kettering’s 3F8 antibody (research also began in 1985) and 14G1,14G2b, and 14G2a antibodies. The ch14.18 chimeric antibody is a “second generation” anti-GD2 antibody, since it has been engineered to be 75% human and 25% mouse in makeup, and why it is labeled chimeric (a “mix” of human and mouse). This greatly reduced the incidence of forming antibodies against the ch14.18. Two “third generation” antibodies that are fully humanized have been developed to date and have been tested in clinical trials:
- St Jude’s hu14.18K322A, in a phase 1 study now for neuroblastoma and melanoma, and given without cytokines
- hu14.18-IL2, a fusion protein where the cytokine IL2 is attached to the antibody (in phase 2 study now for melanoma)
The hu14.18-IL2 antibody has already shown significant efficacy in a phase II study for relapsed and refractory neuroblastoma with results just published in October 4, 2010 issue of the Journal of Clinical Oncology.
Plans are underway now for both ch14.18 and hu14.18-IL2 to be used in further clinical trials in combination with other promising agents for relapsed/refractory neuroblastoma and these trials will begin accruing at COG institutions in 2011.
More to the story
Ironically, this pivotal phase III ch14.18 trial that showed such a dramatic improvement to survival had some difficulty accruing. It is interesting to note that the other recent phase III studies all accrued patients at a relatively even pace (~90-100 patients per year) with the exception of this ch14.18 antibody study:
- CCG-3891 (1991 – 1996) double randomization of transplant and cis-retinoic acid accrued 539 over 6 years or ~ 90 per year 
- COG-A3973 (2001 – 2006) randomization for purge vs no purge of stem cells for stem cell transplant accrued 489 over 5 years or ~ 98 per year 
- COG-ANBL0032 (2001 – 2009) randomization of ch14.18 vs no ch14.18 accrued 226 over 7.5 years or ~ 30 per year 
- COG-ANBL0532 (2007 – 2012) randomization of single vs tandem transplant is accruing on schedule (should be complete by fall 2012) at 495 over 5 years or ~ 99 per year 
The striking fact is that if the early analysis had not revealed a significant difference in outcome, accruing at this rate this trial might have been ongoing until 2014.
Medical ethics, trial design, and real children
With success also comes inevitable heart ache. Hindsight can be a bitter pill to swallow. It is impossible to forget the children who did not receive the antibody and had increased chance of relapse as a result. By the time 2 years elapsed from randomization, 38/113 children had relapsed after receiving the antibody, but 61 children had relapsed after receiving no antibody, an excess incidence of relapse in 23 children. Was it really necessary to randomize the antibody? If it was a promising treatment why was it not just given to everyone?
There are no easy answers to this fair and difficult question. While there were high hopes the ch14.18 antibody given with two cytokines would help, no one really knew if it would make a difference in survival. After all, in 2004 the German study group (GPOH) had published their retrospective findings that the ch14.18/CHO antibody (made with hamsters instead of mice, and given without cytokines) made no difference in survival when groups were compared from GPOH NB90 and NB97 protocols.
A perfect example of this very quandary was played out with neuroblastoma not long ago. A method was devised in the early 1990s to purge stem cells of neuroblastoma with monoclonal antibodies (of all things) and magnetic beads. The purged stem cells could then be frozen and returned to children after high-dose (myeloablative) chemotherapy. This idea made so much sense: why not clean up the stem cells first and remove the risk of re-infusing the child with NB cells?
Fast forward to the negative results of a very costly and lengthy phase III study — purging had made no difference at all in the survival of high-risk NB children. These results were presented at the 2007 ASCO meeting, but are still not published to date. 
So what are the implications? The purging costs upwards of $30,000 per child. It also wastes 50% or more of the stem cells in the process. Knowing that this expensive, wasteful process is not needed is a very important finding. A similar finding could have been in store for ch14.18 with cytokines. Randomizing is not necessary when a dramatic and consistent response results from a treatment. Not every child responded to ch14.18 treatment in earlier studies, so efficacy had to be proven before it could become a standard treatment. After all, 5 months of ch14.18 treatment with cytokines is a very expensive and complex ordeal, and children are required to spend up to 7 additional weeks in the hospital for this intensive treatment.
In the midst of the celebration over this genuine breakthrough, it is nevertheless heartbreaking to realize that a total of 99 children out of the 226 (both groups) had relapsed by two years — or 44%. It is poignant to note that each of the researchers interviewed about this remarkable study also made the comment “We must do better.” There is an impressive array of researchers and clinicians who have dedicated their entire careers to pushing that sad high-risk neuroblastoma survival curve upward. They see the faces of the children who have been lost along the way in those curves too.
Costly development and production
Developing an antibody (a biopharmaceutical) is far more complex that developing a drug. Cost of production and additional regulatory requirements make this an expensive endeavor. For example, $8 million of 2009 stimulus funds were awarded December 2009 to SAIC-Frederick (NCI research partner) to produce a two year supply of ch14.18:
NCI, through the BDP [Biopharmaceutical Development Program], is to deliver sufficient number of vials of finished product to treat all neuroblastoma patients for whom antibody Ch14.18 has become the clinical standard of care. This 2-year interval for NCI production can be used as a transition to licensing and commercial production. In addition, for the Cancer Immunotherapy Network, the NCI, through the BDP, will develop and supply vials of agents of great interest of the extramural community for further clinical investigation.
Transitioning is currently underway for United Therapeutics to begin producing ch14.18, and complete the FDA registration process. Keeping an eye on further use in melanoma is of interest since that will potentially make ch14.18 a more profitable product for United Therapeutics.
Implications for Europe
At ANR (Advances in Neuroblastoma Research) in 2008 and 2010 and at SIOP 2009 the German group (GPOH) reported that after longer follow-up, the ch14.18/CHO treatment might prevent late relapses. The GPOH is planning to reintroduce ch14.18/CHO treatment. The large SIOP trial SIOP-EUROPE-HR-NBL-1 opened in 2002 and had planned to randomize ch14.18 but since the results of the COG study, the SIOP study was amended to give all eligible children ch14.18 with or without subcutaneous IL2. There is such a great body of evidence showing that GM-CSF is an essential part of this treatment, hopefully the regulatory hurdles will be quickly resolved and children in Europe will soon have the opportunity to get this better treatment regimen. See John Roger’s ANR report for more on this very important subject.
2. Cancer Research 49, 2857-2861, June 1, 1989. Functional Properties and Effect on Growth Suppression of Human Neuroblastoma Tumors by Isotype Switch Variants of Monoclonal Antiganglioside GD2 Antibody 14.18
4. J Clin Oncol. 2009 Mar 1;27(7):1007-13. Long-Term Results for Children With High-Risk Neuroblastoma Treated on a Randomized Trial of Myeloablative Therapy Followed by 13-cis-Retinoic Acid: A Children’s Oncology Group Study
5. Journal of Clinical Oncology, 2007 ASCO Annual Meeting Proceedings Part I. Vol 25, No. 18S (June 20 Supplement), 2007: 9505 Response and toxicity to a dose-intensive multi-agent chemotherapy induction regimen for high risk neuroblastoma (HR-NB): A Children’s Oncology Group (COG A3973) study
6. N Engl J Med. 2010 Sep 30;363(14):1324-34. Anti-GD2 antibody with GM-CSF, interleukin-2, and isotretinoin for neuroblastoma.
7. Correspondence with investigators
9. Journal of Clinical Oncology, 2007 ASCO Annual Meeting Proceedings Part I. Vol 25, No. 18S (June 20 Supplement), 2007: 9505 Response and toxicity to a dose-intensive multi-agent chemotherapy induction regimen for high risk neuroblastoma (HR-NB): A Children’s Oncology Group (COG A3973) study
10. Clinical Cancer Research August 1997 3; 1277 Phase IB trial of chimeric antidisialoganglioside antibody plus interleukin 2 for melanoma patients.