This is the story of our low-temperature-sensitive liposome (LTSL) (Needham et al. 2000), subsequently called ThermoDox^® by the company Celsion Corporation which licensed it from Duke University in November 1999. It is an account of “lessons learned,” as told by me, as I saw it, experienced it, and lived it, and so much of this will be written in the first person. Everyone has their own story. I, myself, Celsion, and ThermoDox^® have ours. As Bernard Malamud says in his book, The Natural, “We have two lives, the one we learn with, and the life we live after that” (Malamud 1952). This certainly applies to me and Celsion. So I write this not for Celsion, or for Duke, but for the next generation of inventors, carers, and entrepreneurs, who have a life to learn with and perhaps have yet to make their own mistakes or achieve their own real and verifiable successes. This chapter contains some of what you might have in store. I encourage you to heed these “lessons learned.” See if you can benefit from the positive results and avoid the ones that, let’s say, made life more difficult. At least be aware of what could happen if you, as the inventor, trust anyone else with your precious invention, which you will have to do. I know this is a long and detailed story, but please try and make it all the way to, what will hopefully be, a happy ending.

For example, these early applications were hampered by the now relatively well-understood opsonization phenomenon that labels most nanoparticles in the bloodstream as foreign for removal by the reticuloendothelial system (RES) (Roerdink et al. 1981). In their early formulations (Gregoriadis et al. 1971, 1974), Gregoriadis and colleagues had used a composition of 40 μmol of phosphatidylcholine, 11.4 μmol of cholesterol, and 5.7 μmol of the negatively charged phosphatidic acid (molar ratio 7:2:1), giving liposome that was 10 mol% negatively charged. When Alec Bangham took me out to a Chinese dinner on the eve of my giving his 80th Birthday Lecture at Gregory’s 2001 5th International Conference on Liposome Advances, in London (Needham 2001a), he told me a story from those early days. Alec had explicitly advised that the initial strategies to make the liposomes “repulsive,” by including negative charge, which was thought to be electrostatically stabilizing, would conversely only result in liposomes going to the RES faster than ever. This has since been proven (Allen et al. 1988) and is fairly well understood (Liu et al. 1995) in serum and serum-free media (Rothkopf et al. 2005). As we now all appreciate, this opsonization to Kupffer cell removal of the liposome in the bloodstream was solved by applying earlier work on the PEGylation of proteins (Beauchamp et al. 1983), and the invention of the so-called “stealth” liposome (Allen 1989), with its sterically stabilizing polymer (Allen and Gabizon 1990; Klibanov et al. 1990), forming the basis for Doxil^® (see Barenholz’s review and references therein, Barenholz 2012). There is also a lesser-known mechanism of a mechanical effect of high cholesterol content in PC lipids or sphingomyelin (Kim and Needham 2001; Needham and Nunn 1990), which gives a similar long circulation half-life (Lasic and Needham 1995), used to good effect (slow leakage and long circulation) in the vincristine liposome (Boman et al. 1993, 1994; Kanter et al. 1994; Krishna et al. 2001; Webb et al. 1995, 1996).

I mention these early studies and “lessons learned” because newbies (and “oldbies”) to the field of nanomedicine would do well to read through some of these early reports and seek to inform their own current research strategies. “Nanomedicine” does seem to know all about PEG-derived steric stabilization, but is it as well appreciated in the nanomedicine literature that a tight lipid-surface is also not well opsonized and that an amount of negative charge of only 3 mol% or greater (Allen et al. 1988) can send your precious nanoparticle rapidly to its RES fate?

As Frank Szoka (who entered the field in its “golden age”; Szoka and Papahadjopoulos 1978), famous for ranting in an editorial Commentary: Rantosomes and Ravosomes (Szoka 1998), reminded us, quoting the philosopher, essayist, and poet, George Santayana: “Those who cannot remember the past are condemned to repeat it.” The example Frank chose to illustrate this was liposomes carrying anti-HIV agents. He points out that

The difficulties with the approach could be anticipated based upon the early liposomal anti-cancer drug delivery literature. The failure of those working in the field is not because they repeated previously published studies. But rather because they did not translate the lessons from the early studies into a therapeutic strategy that complemented the pathophysiology of the disease, with the pharmacological aspects of the drug and the delivery properties of the liposome.

Thus, to reiterate in this editorial, Frank’s amendment to the Santayana quote was

Those who cannot translate the lessons of the past are condemned to repeat it.

Frank, correct me if I am wrong, but basically what I think you were saying is

Those of you who do not read the literature are condemned to repeat it! And guess what? We have already done it, and we are tired of hearing about your ‘new work’ at meetings, or seeing what you think is new and exciting in your grant proposals, that we have to take our precious time to read through, review, and reject.

Back then, the (liposome) world was certainly smaller, and as Szoka also points out, many of the potential uses of liposomes in drug delivery that have since come to be were discussed in those two 1976 review articles by Gregory Gregoriadis (1976). With only a few people involved, “discussions raged over materials, mechanisms, models, methods and structures” at the handful of liposome/drug carrier meetings that everyone attended.

Many lessons were, in fact, learned back then, face-to-face, between researchers during the meetings. But today, even with (and perhaps because of) easy access to the (voluminous) literature, this face-to-face contact is easily lost, and so the repetition of research and, indeed, the mistakes, still go on today. Frank, who rants on average about every 15 years, has another one. This time, it is directed at all nanomedicines and the people who research and develop them. Entitled “Cancer nanomedicines: So many papers and so few drugs!” (Venditto and Szoka 2013), folks would do well to heed it, as this most comprehensive review identifies a timeline to nanomedicine anticancer drug approval using the business model of inventors, innovators, and imitators. Unfortunately for today’s researchers who are still involved in “liposomes,” there is a lot of literature, some 48,772 publications currently listed (February 28, 2015) on a PubMed search for “liposome.” And incredibly, 339,246 patents that concern “liposome” at free patents online. “Lessons learned” from liposomes, like Chezy Barenholz’s landmark paper Doxil^®—the first FDA-approved nanodrug (Barenholz 2012), will undoubtedly apply to all nanomedicines being researched and developed today. With only 9422 papers found on PubMed for a search of “nanomedicine,” it is likely that “your precious nanomedicine” shares many of the problems that were already figured out by the liposomologists, maybe before you were born or, at least, while you were still in grade school. So where would a company turn to, in order to get the most expert and up-to-date knowledge about their licensed product? Would they sift through the 48,772 papers and pay a law firm $gazillions to search through the 339,246 patents? No. Maybe try asking the inventor, and first see what he or she has to say (and more about that relationship later).

Early commercialization efforts were considerable (as well as contentious; Free Library 2014; NeXstar Pharmaceuticals 1997). From my perspective, there were four principal liposome companies: three were established in the United States: Liposome Technology Inc. (LTI) in Menlo Park, California; The Liposome Company Inc. in Princeton, New Jersey; NeXstar, based in San Dimas, California; and Inex in Vancouver, Canada, first as a spin-off from TLC and then in its own right. These were the main entities that, for me, paved the way in mainly anticancer liposomal therapeutics, took relatively similar drug delivery ideas, and brought them through the development and translation to commercialization. Two companies focused on liposomal doxorubicin, LTI (Doxil^®) and The Liposome Company (Evacet^TM and Myocet^®), NeXstar went with daunorubicin (DaunoXome^®) and also successfully developed amphotericin B (AmBisome^®), and Inex developed liposomal vincristine (Marqibo^®).

With academic researchers involved at varying levels of invention, innovation, and continued support, there is a whole host of very experienced individuals (still) around the conference, industrial, and grant-review world, whom students, postdocs, and young faculty new to the field can call on for advice and informal consultation. So, if you see Chezy Barenholz (Barenholz 2012), Frank Martin, or Martin Woodle et al. (Barenholz and Haran 1994; Papahadjopoulos and Skoza 1980; Woodle et al. 1989) ex-LTI-Sequus; Andy Janoff et al. (Janoff et al. 1989, 1996; Mayer et al. 1997) ex-TLC; Pieter Cullis, Lawrence Meyer, Marcel Bally, or Murray Webb et al. (Bally et al. 1997; Webb et al. 1996; Wheeler et al. 1999), ex-Inex; or Gary Fujii, Bill Ernst, Jill Adler-Moore, or Su-Min Chang et al. (Adler-Moore and Chiang 1997; Adler-Moore and Ernst 1997; Proffitt et al. 1999; Schmidt and Fujii 1998) ex-NeXstar; or very knowledgeable and outspoken independents, like Frank Szoka (Papahadjopoulos and Skoza 1980), Theresa Allen (Allen 1989; Allen and Gabizon 1990), or Valdimir Torchilin (Torchilin et al. 2006), or me and Dewhirst, at a meeting or seminar, we are all inventors of some of the earliest or later patents and a host of peer-reviewed papers that have contributed to those 48,772 publications. You might ask us one or more of the questions I am hoping to address in this chapter: What was the original observation that motivated the product? How did you or your colleagues have it? What was the question you tried to answer? What was the scientific hypothesis that you and your colleagues addressed? What funds were available for initial development and how did you get them? How did you develop it, including preclinical animal studies, and what were the results? What collaborators did you work with? What was the initial licensing deal and how was it structured and with whom? How did you manage the scale-up of manufacturing? What particular disease did you decide to treat and why? What did the Food and Drug Administration (FDA) require you to write in order to obtain an Investigational New Drug (IND) application? What funds did you have to raise, in order to carry out the IND application and how did you get them? How did you manage the Phase 1 trials, and what was the result? How did you recruit the clinicians and the sites to carry out the trials? Were there any unexpected problems implementing the protocols? Was that next stage a Phase 2 or did you go straight to Phase 3 human clinical trials? What was the next documentation you had to provide to the FDA and how long did it take to move to this next stage? What was the protocol? How many patients and sites did this Phase 3 involve, and how was it all managed? How did it turn out? Are you still in business? If so, what are your future plans for your drug delivery technology? In your view, what’s next? What’s the next big thing in advanced drug delivery and why? The answers are all available if you would just ask.

Invented in 1996 (Needham 2004), the LTSL is now in Phase 3 human clinical trials for liver cancer. This chapter will describe the engineering design of the LTSL (Needham 2013) and the licensing and clinical testing that is moving the invention toward commercialization. LTSLs, in conjunction with local mild hyperthermia (HT), can release drug within seconds of entering the warmed tumor vasculature. The released, free drug diffuses into the tumor interstitium, reaching its nucleus target with greater penetration distance, and to much higher concentrations, than those achievable by either free drug or the more traditional long-circulating liposome formulations (Manzoor et al. 2012). Intravascular drug release provides a mechanism to increase both the time that tumor cells are exposed to maximum drug levels and the penetration distance achievable by drug diffusion. This establishes a new paradigm in drug delivery: rapidly triggered drug release in the tumor bloodstream, saturating neoplastic cells, as well as endothelia, pericytes, and stroma, with the anticancer drug.

This chapter will attempt to describe the process, the pitfalls, the good, the bad, and the ugly of translating what might be a good research idea through initial funding, development, and preclinical and clinical trials. Whether unique or not (and we suspect the issues highlighted in our particular case are probably quite ubiquitous), I will use, as the main example, the invention, development, clinical testing, and intended commercialization of the thermal-sensitive liposome, subsequently named ThermoDox^®. Additionally, an update of the recent progress in human clinical trials will be given, including Celsion’s Phase 3 for liver cancer, Phase 2 recurrent chest wall (RCW) cancer, and their new trials in metastatic liver cancer, ovarian cancer, pancreatic cancer, breast cancer, and glioblastoma, including the adaptation of high-frequency ultrasound (HiFu) as the source of targeted mild HT. I will say up front that the current Celsion administration should be commended for their pioneering work in taking on this invention, initiating this very impressive series of trials, with multiple sites, in different countries, and in particular for creating and carrying out what actually is the largest randomized double-blind trial ever with radio-frequency ablation (RFA). For example, following on from what was learned in the Phase 3 HEAT study on primary liver cancer, Celsion’s new OPTIMA trial is currently enrolling, with the first patient enrolled Q3-2014. It will include approximately 550 patients in up to 100 sites in North America, Europe, China, and the Asia-Pacific region.

Kudos to Celsion and in particular Mike Tardugno (CEO) and Nick Borys (CMO); it’s not an easy task visiting all these clinical sites and spending multiple days on airplanes.

Following the licensing of my (Oops, sorry, no, it’s not mine!) Duke’s invention by Celsion Corporation in November 1999, ThermoDox^® has now, just over 15 years later, completed a Phase 3 trial in 701 patients with primary liver cancer, where ThermoDox^® + RFA was trialed against RFA alone (Poon and Borys 2009; Wood et al. 2012). There were certainly lessons learned in the science (and all that is published in several reviews; Landon et al. 2011; Needham 2013; Needham and Dewhirst 2001, 2012) and the seminal paper on drug accumulation and penetration by Manzoor et al. (2012). But why we think there are real lessons to be learned in the development and commercialization from all this in particular, is the fact that when the 701-patient Phase 3 primary liver cancer trial results were unwrapped by Mike Tardugno and colleagues, on January 31, 2012 (Tardugno 2013), the headline read as follows:

ThermoDox^® failed to meet its primary endpoint of better progression free survival than the heating modality (RFA) alone.

When we decided to develop the LTSL technology, we simply inherited the “stealth” liposome concept and the cytotoxic drug, Dox, which had already been used in liposomal delivery. And so, as a prelude to discussing actual ThermoDox^®, it is worth a relatively brief (not exhaustive) review of what we knew about liposomes and Dox in 1995, highlighting a few comparisons, even a few lessons learned as we go, and exactly what it was about liposome technology and its performance at the time that motivated the LTSL invention and its development.

Dox (also known as hydroxydaunorubicin and hydroxydaunomycin; trade name, Adriamycin^®) is an anthracycline antibiotic discovered in the early 1960s by the Farmitalia Research Laboratories of Milan. Clinical trials began in the 1960s, and the drug saw success in treating acute leukemia and lymphoma. It was approved for use in the United States in 1974. However, by 1967, it was already being recognized that Dox (and another anthracycline chemotherapeutic, daunorubicin), while active against cancer, could produce fatal cardiac toxicity.

So, quietly, between ourselves, Mark Dewhirst and I had questions in 1993–1994, born out of our own preclinical animal studies: “could Doxil^® really deliver enough drug to human tumors to achieve significant efficacy?”

Attempting to improve Doxil’s efficacy by using HT to increase the vascular permeability, the Hyperthermia Center at Duke also carried out two “Doxil^® + HT” clinical trials in ovarian and breast cancers (Secord et al. 2005; Vujaskovic et al. 2010). Unfortunately, the clinical trial evaluating the combination of Doxil^® + HT in patients with persistent and recurrent ovarian cancer did not demonstrate increased efficacy, compared with prior reports using Doxil^® alone. And in a Phase 1/2 study of paclitaxel in patients with locally advanced breast cancer in the preoperative setting, the pathological complete response (pCR) rate was only 9% for liposomal Dox (Evacet^™) + local HT, although the combined (i.e., paclitaxel + liposomal Dox [Evacet^™] + local HT) pCR rate was 61%.

Although we did not know it at the time, therapeutically all these concerns were, in fact, borne out in subsequent human trials, including a trial that was instrumental in gaining Doxil’s approval for ovarian cancer. A trial for metastatic breast cancer showed that Doxil^® really was no better than free drug alone (O’Brien et al. 2004). That is, even though Dox has a half-life of only 2 minutes and Doxil^® has a half-life of 73.9 hours:

•  The overall survival was comparable with both treatments.

•  Median: PEGylated liposomal Dox 21 months, versus doxorubicin, 22 months; hazard ratio (HR) = 0.94 (95% CI 0.74–1.19).

•  At the time of the analysis, approximately 56% of patients in each group had died. When adjusted for potential imbalances in prognostic variables using the Cox regression analysis, the HR was 0.94 (95% CI 0.75–1.19), similar to the unadjusted treatment HR.

Doxil^® was approved for ovarian cancer (Gordon et al. 2001) in 2005, but again, with little therapeutic benefit, when compared to topotecan:

•  Time to progression (TTP): Doxil^®, 4.1 months, and topotecan, 4.2 months

•  Overall median survival: Doxil^®, 14.4 months, and topotecan, 13.7 months

•  18% reduction in risk of death (hazard ratio [HR] = 1.216)

•  Overall tumor response rates: 19.7% (47 patients) in the Doxil^® arm and 17% (40 patients) in the topotecan arm

Compared to topotecan then, improvement in TTP for Doxil^® was 0.1 month, which is only a 2.4% progression-free survival (PFS), and it was approved. TTP was also a measure for ThermoDox^®, and we will see later how it faired (when taken in total, it didn’t fare well), in relation to the FDA-required %PFS stipulated for its approval.

While all this liposomal development was going on, we also knew that tumors could be heated using HT, or at least Mark Dewhirst did. Mark and his colleagues had been working in HT since the early 1980s. In a monograph from 1983 (Oleson and Dewhirst 1983), Oleson and Dewhirst focused on progress at the time regarding “major aspects of the biologic effects of elevated temperatures both in vitro and in vivo, on the physical methods clinically used to produce HT, and on the results of treatment in large animals and humans.” It is well recognized that HT can have at least three different effects on cells:

•  Direct cytotoxicity at elevated temperatures, where the degree of cell killing is both time and temperature dependent

•  Heat sensitization of radiation, where HT reduces cancer cells ability to repair sublethal and potentially lethal radiation damage

•  Synergistic effects with certain drugs, including alkylating agents, nitrosoureas, cisplatinum, bleomycin, and adriamycin, which showed marked synergism above 43°C

Following the pioneering characterization of the red blood cell (RBC) membrane in the 1970s (Evans and Hochmuth 1976), “lipid membrane mechanochemistry” as a field of study for pure lipid bilayers was really initiated in the early 1980s by Evan Evans, when he moved in 1980 from Duke Biomedical Engineering to the University of British Columbia, Vancouver, Canada. But he started endogenously, by checking out nature’s own designs for lipid-based capsules.

While at Duke (1973–1981), Evans had developed and used the micropipette manipulation technique to characterize the mechanical properties of RBCs. Interestingly, the biochemists (Steck 1974) were only just starting to identify the composition and organization of the spectrin membrane cytoskeleton (Bennett 1982) at the same time that Evan was measuring with the micropipette technique (Evans 1973a,b; Evans et al. 1976). Thus, sophisticated mechanical models of the RBC membrane were introduced in the early 1970s by both Skalak (Skalak et al. 1973) and Evans (Evans 1973a). Evans unified a new material concept for the RBC membrane, which provided the capability of large deformations exhibited by normal discocytes as a two-dimensional, incompressible material, and a general stress–strain law was developed for finite deformations (Evans 1973). What followed was a period of intense activity (Evans and Hochmuth 1977, 1978), culminating in the seminal book by Evans and Skalak Mechanics and Thermodynamics of Biomembranes (Evans and Skalak 1980). Thus, for over 50 years, these micropipette techniques have provided the unique ability to apply well-defined stresses for dilation, shear, and bending modes of membrane and cellular deformation. They laid the foundation for similar micromechanical experiments applying well-defined stresses to single-giant unilamellar vesicles (GUVs) that, in turn, eventually generated the “idea” for the LTSL. This is how that happened.

The classic materials engineering approach to either understanding an existing design (like the red or white blood cell) or creating a new design for a nanoparticle drug delivery system like a liposome is the same. It involves as complete an understanding as possible for the composition–structure–property (CSP) relationships of the materials involved in the component design. To be fair, most liposomologists did not have access to the more mechanochemical techniques we were using, and so in contrast to the development of most liposome and nanomedicine formulations, we didn’t just come up with an all-encompassing series of lipid–drug compositions and evaluate them for performance; we came up with one. Of course, this was based on liposomes that had already been developed and tested, especially the stealth version (Allen 1989). But with over 20,000 papers in the literature on liposomes at the time, new innovation can often require a deeper understanding of CSP relationships. And so, by the time Mark Dewhirst asked for a formulation he could “heat and it released drug,” we (and Evans before me) had been studying membrane materials science for over two decades and had already evaluated many of their CSP relationships using the micropipette technique. These micromechanical methods and the resulting data on single-bilayer vesicles have allowed us to characterize the lipid bilayer membrane in both a liquid and a solid state, including the influence of cholesterol on membrane elasticity and tensile strength. It was these kinds of direct measurements of the “two-molecule-thin” material that were, literally, instrumental in helping to characterize liposomal systems and explain much of the processing and performance of the liposome.

Starting in the 1980s, the micropipette technique was adapted and developed by Evans and Kwok (Evans and Kwok 1982; Kowk and Evans 1981) and then established by (Evans and Needham 1987), to study individual GUVs of various lipid compositions. A particularly historical and newly insightful perspective on the mechanics and thermodynamics of lipid biomembranes has been given recently (Evans et al. 2013), which emphasizes “the inherent softness of fluid–lipid biomembranes and the important entropic restrictions that play major roles in the elastic properties of vesicle bilayers.” In particular, the properties of importance for the LTSL were

•  The membrane elastic modulus and other mechanical properties that determine drug loading and retention

•  The nature of its main acyl-melting phase transition, which was the key trigger for drug release

•  The behavior of the gel-phase membrane in shear and in particular its degree of yield shear and shear viscosity, as a result of in-plane shear deformations of membrane-grain structure

•  Molecular exchange with water-soluble and membrane-soluble molecules like lysolipids that modified the membrane to allow it to have such a rapid release of encapsulated drug

•  The adhesive and repulsive interactions that help maintain stability in blood circulation

Source:  Reproduced from Biophys. J., 73(5), Needham, D., Stoicheva, N., and Zhelev, D.V., Exchange of monooleoylphosphatidylcholine as monomer and micelle with membranes containing poly(ethylene glycol) lipid, 2615–2629. Copyright 1997, with permission from Elsevier.

In the experiment, we first measured a fundamental force of nature, van der Waals attraction, limited at PC membrane surfaces by a very short-range hydration repulsion. We then started adding in all the other potentials, as described in Evans and Needham (1987). By adding in charged lipids, this attraction could eventually be overcome by ≈5 mol% phosphatidylserine, which fitted according to the well-known DLVO theory for colloid stability. For neutral vesicles, adding extraneous nonadsorbing polymers, like PEG or dextrans, generated large attractive stresses and highly measured adhesion energies, which again were well modeled by depletion–flocculation theory (Evans and Needham 1988). These attractive potentials were so large they could overcome even mutual negative repulsions at the interfaces of up to 30 mol% charged lipids. Finally, on this baseline, the inclusion of PEG lipids caused separations of the membranes and overcame all long-range colloidal forces and generated the “stealth” effect, as described in more detail with Dan Lasic (Lasic and Needham, 1995).

Although not part of the LTSL story, we (with Dorris Noppl, a student exchange from Germany) used the same two-vesicle experiment to observe and measure receptor-mediated adhesion (Needham and Kim 2000; Noppl-Simson and Needham 1996), an effect that could have some bearing on ligand targeting.

To summarize then, and as described more fully in another recent review (Bagatolli and Needham 2014), these experiments have characterized a two-molecule-thick membrane:

•  It is a soft elastic material with a compressibility somewhere between that of a bulk liquid and a gas.

•  It is stiffened considerably by the inclusion of cholesterol to levels equivalent to polyethylene.

•  It is permeable to water in relation to its compliance.

•  It displays a 25% change in area when taken through its main acyl chain freezing transition.

•  As a gel-phase material, it shows the yield shear and shear viscosity of a Bingham plastic.

•  It can exchange small amounts of other lipids and surfactants with its surrounding milieu.

When manipulated in pairs, lipid vesicles have also been found to be subject to the same range of attractive and repulsive colloidal interactions as many other colloidal particles, including

•  Van der Waals attraction, limited by a very-short-range hydration repulsion and the variable-range power law of electrostatic repulsion

•  Steric repulsive barriers due to the presence of bound aqueous polymers like PEG, incorporated as, for example, PEG lipid

•  Depletion and flocculation by water-soluble polymers like PEG and dextrans when free in surrounding solution

•  Intimate mixing (fusion) when manipulated into contacts that reduce the hydration barrier and allow membrane–membrane fusion

DMPC is not much smaller (2× (CH₂)) than DPPC, the lipid used in ThermoDox^®. And so, even in 1983–1987 (over 10 years before I invented the LTSL), these thermomechanic studies of GUVs were solidifying in my mind exactly what happens when a liposome is taken through a phase transition by changing the temperature, in terms of its physical behavior, the membrane area change, the broadening by a second lipid or cholesterol component, and the shear and viscosity as a result of sliding grain boundaries in the membrane surface. All that was needed now to provide the “light bulb” of the LTSL idea was a fundamental understanding of what happens when a lipid vesicle is exposed to a soluble lipid that can partition into the membrane and can be washed out again, upon changing the bathing medium.