Twenty years of research in melanoma therapy–From “nothing works” to cures: A personal account

In the late 1990s and early 2000s the mood in the melanoma field was grim. “Nothing works,” said our oncology colleague Lynn Schuchter, after the third large Phase III trial in immune therapy (a MAGE3 trial) failed. Don Morton's Bacillus Calmette-Guérin (BCG) trial had also just failed, and he was truly disappointed (Eilber et al., 1976; Morton et al., 1974). There were no alternatives, no hope. The melanoma research field was small, underfunded, and isolated. While oncologists, surgeons, pathologists, and epidemiologists continued spirited debates about their newest findings in specialized meetings, the melanoma experimental researchers were few and widely scattered. Prior to the founding of the Society for Melanoma Research (SMR), melanoma researchers did not have an intellectual home. There was no organized pipeline for attracting young researchers, there were no tissue banks, no databases, and the field lacked animal models beyond a few transplantable tumors such as the B16 model or nude mouse xenograft models. The incidence of melanoma had been rising since the 1950s at an alarming rate of 2%–5%/year. Treatment of advanced disease had not improved in the past 30+ years and the failures of the latest large clinical trials hammered down the point that melanoma ranked among those cancers with the lowest 5-year survival rates, almost on par with pancreatic cancer or glioma.

The field could build on progress, which demonstrated an immune response could be activated in melanoma patients. Monoclonal antibodies had helped to define ~200 melanoma-associated markers, mostly cell surface receptors used for adhesion or growth signaling. However, those monoclonal antibodies defied all attempts to use them as “magic bullets” for therapy. The first oncogene (NRAS) in melanoma was initially defined by Anthony Albino, and colleagues at MSKCC (Albino et al., 1984), but NRAS continues to defy any therapeutic targeting to this day. [Correction added on 19 October 2023, after first online publication: In the previous sentence, “The first oncogene (NRAS) had been” was changed to “The first oncogene (NRAS) in melanoma was initially”.]

In June 2002, Barbara Weber from the University of Pennsylvania called: “Tomorrow there is a Nature paper from Mike Stratton's lab at the Sanger Institute, coming out on a new oncogene that will transform the field.” Andy Futreal, Michael Stratton, and colleagues had conducted a tour-de-force in sequencing cancer cell lines of different origin. A point mutation in the BRAF gene was found predominantly in melanoma (Davies et al., 2002). A new research era began for experimental and clinical researchers alike. Initially, the field was skeptical because the MAPK pathway is activated in nearly all melanomas regardless of the mutational profile. BRAF^V599E became BRAF^V600E, conquered the field, and captured the imagination of many cancer researchers. However, in 2007, when we submitted a Program Project on targeting BRAF^V600E the reviewers said, “BRAF? Why develop a program on it?” One year later, when we submitted the revised grant, the reviewers were still not convinced of the significance of BRAF mutations for melanoma. Only two years later, when the first clinical data became known, reviewers were enthusiastic. Five years later, at renewal time, the National Cancer Institute (NCI) told us to forget BRAF as a theme because they were already funding three more programs and several single investigator grants, all on BRAF. We just slightly modified the theme, and we were fine. BRAF^V600E had conquered the field. Richard Marais showed us that BRAF^V600E is 800-fold more active than its wild-type counterpart, which convinced even the few remaining skeptics of the significance of this oncogene in melanoma. The real breakthrough for clinical application came from a small biotech company, Plexxikon, which had used structure-based design to develop mutation-specific inhibitors. Through our colleague Keith Flaherty, we obtained and tested PLX4720, a tool compound that Gideon Bollag (then at Plexxikon), provided to many academic researchers (Tsai et al., 2008). The openness of Plexxikon to academic collaborations greatly accelerated progress in the research field. Big pharmaceutical companies like Bayer with its multi-kinase inhibitor sorafenib shielded its drugs from academics. Bayer lost out on critical input on the strength and weaknesses of this drug. It did not take long and sorafenib was dropped from the clinical arsenal and largely ignored by researchers in the melanoma field.

The initial trial of PLX4032 (vemurafenib), led by Keith Flaherty, had to be halted because of poor half-life in vivo. Plexxikon had sold its main share of the drug to Roche. Big pharma demonstrated their incredible resourcefulness and the company put an army of medicinal chemists to work to develop a better formulation of the drug and the trial continued within a few months. Vemurafenib generated big excitement because in some patients the tumors “melted away.”

The first mouse genetic models became available, particularly the BRAF^V600E/PTEN^−/− model developed by Martin McMahon and Marcus Bosenberg (Dankort et al., 2009), but this and other models (Dhomen et al., 2009; Pérez-Guijarro et al., 2017) were too late to direct clinical trials with new strategies. Instead, the clinical community raced with a series of clinical studies allowing multi-institutional trials at breath neck speed and efficiency. MEK inhibition was added to BRAF and clinical trials showed that MEK inhibitors in combination with BRAF inhibitors decreased toxicities and increased efficacy. Two additional pairs of BRAF/MEK inhibitors were added; they were traded several times between companies with Novartis as the main player.

Despite the excitement and promise of the new therapies, it did not take long to realize that drug resistance would be a daunting challenge. Who has not seen the famous photos of the trunk of a male patient before he was treated with a BRAF inhibitor, during therapy, and after relapse? These pictures taken by Nikhil Wagle, then a clinical and postdoctoral fellow in the lab of Levi Garraway are likely the most cited photographs in the biomedical field (Wagle et al., 2011). These images graphically illustrate the incredible power of BRAF inhibition but also the devastation for both patients and their families when the tumor roared back. Shortly after the approval of vemurafenib, several groups, including ours, identified various mechanisms of BRAF inhibitor resistance (Johannessen et al., 2010; Nazarian et al., 2010; Villanueva et al., 2010). Other groups quickly added more than a dozen mechanisms of resistance, which have likely doubled to this day.

Since the early stunning success of BRAF targeting, the field hit a glass ceiling and progress has been slow. At an SMR meeting in 2009 or 2010, I made a bet with Michael Lotze, that targeted therapies would help more patients than immune therapy; by 2013 it became clear that immune therapy was the obvious winner due to more sustained responses.

In the early years of targeted therapies (2006–2009), SMR became the galvanizing point between experimental researchers and clinicians, each group benefitting from the input of the others. New discoveries seemed to indicate we could overcome all obstacles including resistance, and then move to cures. This was too optimistic, but the spirit of “we are in this together” never left the melanoma research community. SMR meetings in the Netherlands (Noordwijk), New York, and Boston were exciting because of all the new findings that rained onto the participants and made them giddy with prospects for more.

Immunologists are lucky. Even the poor B16 mouse model gives them valuable information, if the appropriate immunological checkpoints are functional. The model is useless for almost all biological investigations because it is a poor match to the human disease. Jim Allison and many others elegantly demonstrated that checkpoint inhibition can take the breaks off the immune system, which now can effectively target the tumor cells (Huang & Zappasodi, 2022; Leach et al., 1996). Blocking CTLA4 with a monoclonal antibody in the mouse model was very effective in leading rapidly to a clinical trial led by Jedd Wolchok, then at MSKCC; a new era had begun. The initial trial started humbly because blocking CTLA4 is hard on patients leading to Grade-3 and -4 toxicities. Responses came very slowly, much slower than after targeted therapy. The initial patients were dismissed as “nonresponders.” They generated big surprises when, months later, their tumors had shrunk significantly. PD-1 became the second checkpoint that could be successfully blocked with a monoclonal antibody. Clinical responses were more dramatic and faster than with anti-CTLA4, and the drug was less toxic. The combination of both antibodies was even more effective. Checkpoint inhibitors began their victory rally, and they became the treatment of choice for melanoma as first-line therapy in all patients except some rare melanomas such as acral, uveal, and mucosal melanomas. Long-term responses are close to 50%, leading to a drastic reduction of death from melanoma, even if the incidence of the disease continues to (slightly) rise each year. Melanoma has become the leader in immunological research among all cancers, and all major cancers emulate the experience in melanoma. The discovery of anti-LAG-3 as a new checkpoint inhibitor that is effective in combination, at least with anti-PD-1, adds to the mantra of melanoma as a leader in immune therapy. This perception of success has a downside as reviewers and funding agencies ask, “Why do you want to invest more in melanoma when you already get all these cures?” The cautious researchers talk the rousing success of melanoma therapy down and point to patients who relapse and develop resistance to both immune and targeted therapies.

The past 15 years were highly exciting in melanoma research, and SMR was in the middle of the meteoric rise for melanoma research. BRAF/MEK inhibition remains for the 45%–50% of melanoma patients with BRAF^V600E/K mutations a serious/prime therapeutic option since ~80% of patients respond and 25% show long-term complete responses. Even patients with brain metastases respond, albeit they commonly relapse. Unfortunately, patients who relapse after BRAF/MEK inhibition, most often have also relapsed after checkpoint inhibitor therapy, cannot yet being offered a standard second-line therapy. Here, the experimental researchers are being challenged to lead the way for new strategies. Chris Marine and colleagues completed the first in-depth study on resistant cells using single cell RNA sequencing, but a unifying concept remains elusive (Rambow et al., 2018). In addition, currently there are no standard therapies for NRAS mutant melanoma or any of less frequent genetic mutations. Rare melanomas (acral, uveal, and mucosal) continue to be very difficult to treat and in pressing need of more research.

For immune therapy the overall score card is more impressive. Almost half (40%–45%) of Stage IV melanoma patients treated with checkpoint inhibitors, optimally with a combination of PD-1 and CTLA4 inhibitors, are apparently cured, or at least deemed long-term responders. LAG-3 inhibition may replace CTLA4 in the combination with PD-1 inhibitors, but more research must be done. Even more impressive is the shift to treating early disease. Neoadjuvant therapy of Stages II and III melanoma is a clear winner and will dramatically decrease the death rate of melanoma patients. Moving from late-stage to early-stage treatment is of great benefit for the long-term prospect of controlling melanoma progression. Combining immune with targeted therapies is a clear logic extension. However, the clinical trials have shown disappointing results. Can experimental studies have an impact on the recent rapid advances in immune therapy? For this, we would need better models to mimic the human disease (Patton et al., 2021). Immune therapy is currently expanding to “classic” adoptive cell therapy, which may soon be approved, but the approach is incredible involved and expensive. There is a long list of approaches that have not yet shown encouraging results such as CAR T cells. Major vaccine trials are currently ongoing with uncertain outcome. Yet, the immune therapy field remains highly dynamic and may yield new insights at every annual SMR meeting, which attracts both clinical and experimental scientists. Since clinical and immunological scientists have their own intellectual homes, SMR remains critical for the biologists and those who cross both signaling and immune therapies.

Despite tremendous progress and excitement in the field, many questions remain. What are the options for the 60% of patients who relapse after immune therapy? Why do combinations of targeted and immune therapies not act synergistically? Why do the rare melanomas not respond to immune therapy? How can we convert cold tumors into hot? Why have we failed translating inhibitors for PI3K, AKT, RTKs, or inhibitors for growth factor signaling into effective melanoma treatments? How far or deep do we have to reach to find common mechanisms of resistance? Do we have to go beyond the most obvious signaling pathways and search deeper in the cells' ability to survive and thrive despite the drug challenges? Are there resistant cells present prior to any therapy, and how are these different from cells adapting to drugs and immune attack? Would targeting epigenetic or metabolic pathways offer a more effective approach to combat drug resistance?

Each question will likely require complex answers. Apparently, progress nominating new targets and new approaches has slowed for signaling inhibitors. How can we increase the number of real breakthroughs? Maybe not one or two major Eureka moments but gradual, steady progress on many different fronts. Immunologists still feel bullish, but one can easily predict future struggles. For the past 40 years immunologist have focused on T cells with less emphasis on the study of other immune cells and why many can have dual functions as stimulators or inhibitors of tumor progression. The immunology field has largely ignored the tumor cell and its intricate signaling behavior that contributes to the elusive nature of the malignant cells.

The answer for both fields is similar: we must start “listening” to cells, that is the malignant cells and all surrounding tumor associated host cells and the relevant normal host cells in the periphery. Thanks to advances in omics approaches, global gene expression and protein profiles are now available. We can even analyze cells at the single-cell level. Increasingly, through spatial transcriptomics and proteomics, and multiplex immune histochemistry, we can dissect each cell within a tumor or an organ in the tissue context, capturing a realistic read-out of the status quo of a cell. As melanoma cells are highly dynamic and can rapidly adopt different states in response to therapy and conditions in their microenvironment, frequent sampling is necessary to develop a comprehensive map and true picture of each tumor. Of course, patients' lesions provide the most realistic read-out. However, each sampling is a “picture of the tumor frozen in time.” Serial biopsies of solid tumors are essential but difficult, if not impossible, to obtain. Noninvasive imaging technologies are outstanding, but their resolution will likely be of little help to study the dynamics of drug resistance. The logistic and financial challenges are enormous. Interpretation of large data sets from omics is time consuming, very costly, and requires bioinformatics expertise. Our laboratory alone has >80 TB of data to manage. How can we cope with this mountain of data and how can this information guide us in developing new therapeutic strategies that go beyond the current selection of “low-hanging fruit”?

While any material from patients is the best subject for investigation, it is finite and thus difficult to test hypotheses and validate genomic analyses. While new generations of mouse genetic and zebrafish models have significantly contributed to our knowledge of the human disease and how to treat it, each model has severe shortcomings when compared to the human condition. Our own laboratories have attempted to keep human cells from patients, both normal and malignant, alive outside of the human host. Moreover, we are increasingly maintaining cells in a three-dimensional tissue context that is essential for many cellular functions. While far from ideal, the newest models largely reflect human disease and the interactions of the malignant cells with autologous immune cells. We are hopeful we can test hypotheses under experimental conditions that will provide valuable answers for the clinic. The feedback from laboratory to clinic and back needs to be stronger.

How do we continue making progress? The answer for progress in the future is simple: collaboration. Those who gather vast tissue repositories or datasets need to share with those who have unique approaches for analyses and functional studies. Sitting on tissues or descriptive data does not help anyone, including the collectors. We need to develop networks of interested laboratories who freely share data and resources and communicate in a transparent way. We cannot build walls around our laboratories; we need to tear them down. We must develop strategies that extend to each participant, including young investigators, an appropriate and fair share. Likewise, we should apportion ownership to build multidisciplinary networks of scientists. Those networks should be accessible, inclusive, and transparent. Can it be done? We believe so! The NIH has funded priority areas, for example, the NCI's Cancer Moonshot℠ program. Wealthy states like Texas have their own MoonShot programs, which has supported outstanding resources such as tissue banks. Institutions like collaborations on paper, but their push for securing intellectual property rights is often short-sighted and nonproductive, sometimes even hindering. How can a young investigator new to the field navigate these challenges? Here, SMR may need to expand its role as an honest broker, facilitating communication, supporting training of young investigators, and fending off the overreach of academic and nonacademic institutions. SMR was founded to bring laboratory and clinical scientists together. Given the complexity of the field and conundrums with current and future therapies, SMR may need to expand its role to deal with the challenges melanoma researchers face today and most likely in the future. In the past, we had discussions with SMR and patients' foundations on how this could be achieved. That was during the initial science boom years. It is time to revisit this topic and the role(s) of this unique professional society for the benefit of patients and science.

Nothing to report.

The authors have no conflict of interest to declare.