This summer witnessed the approval of a buccally delivered biologic—MannKind's inhaled insulin Afrezza. Earlier in the year, long-acting versions of glucagon-like peptide, Factor VIII and Factor IX reached the market. In the field of short interfering RNA (siRNA) therapeutics, Alynlam's subcutaneously delivered N-acetylgalactosamine (GalNAc) cluster conjugates have achieved potent gene knockdown in the livers of patients—so potent that the technology is now being adopted by antisense developers like Isis and Santaris.

With advances like these, it would be churlish to criticize drug delivery research for a poor track record in marketed products. And yet clinically potent novel approaches like GalNAc-siRNA conjugates remain the exception rather than the rule. Recently approved long-acting versions of macromolecule drugs do not use cutting-edge technology; they use tried-and-tested delivery approaches that are decades old. In the meantime, a whole raft of academic nanotechnology research focuses on developing delivery vehicles that will never see reduction to clinical practice. This matters because the hard problems in macromolecule drug transport—crossing the blood-brain barrier, engaging intracellular targets and accessing solid tumors—will be addressed neither by decades-old technology nor delivery vehicles that pose immunogenicity risks and toxicity concerns. New ideas and new technology are needed. And a meeting convened last month by Roche and Nature Biotechnology suggests that greater understanding of the mechanisms underlying macromolecule transport across cell membranes could pay dividends in the long term.

The drug delivery discipline had its beginnings in the early 1950s. The early years witnessed great strides in creating controlled release formulations, leading to a slew of new pharmaceuticals. In the late 1970s, Bob Langer, working in the laboratory of Judah Folkman at Boston Children's Hospital, first demonstrated extended release of macromolecules using microparticles—at the time a controversial assertion. Elsewhere on the US East Coast, Frank Davis of Rutgers University pioneered the conjugation of polyethylene glycol (PEG) to proteins.

Since that time, all manner of peptides, polymers and lipid microparticles and nanoparticles have been characterized and combined with therapeutic payloads. But in contrast to products from the golden age of oral and transdermal extended release, few drugs resulting from these formulations have ever reached the market. Notable examples include Adagen (the first PEGylated drug), Doxil (liposomal doxorubicin hydrochloride), Enbrel (IgG Fc fused to tumor necrosis factor alpha) and Abraxane (albumin nanoparticle shells loaded with paclitaxel). A handful of similar drugs have followed in their footsteps, but these micro- and nanoparticle technologies have failed to transform macromolecules into effective drugs that can address cytoplasmic targets, cross the blood-brain barrier or penetrate solid tumors in a manner that would radically change clinical practice.

Certain principles have emerged, however. For example, size of a carrier (and polydispersity of a preparation of carrier particles) influences the ease with which a formulation can be injected and whether it will be captured by the reticuloendothelial system. Increased porosity of a microparticle allows penetration of water, and facilitates degradation and payload release. Shape of a drug carrier (spherical, rod, branched or discoid) affects its interaction with phagocytes and the tumbling and rolling dynamics within blood vessels. Surface properties, such as hydrophilicity and charge, are also important: hydrophilic polymers like PEG reduce capture by immune cells; cationic charge (e.g., in cell-penetrating peptides, cyclodextrin or protamine) is thought to buffer the low pH of the endosome and facilitate cytoplasmic escape. And increased deformability of a drug delivery particle may also enhance delivery.

These rules of thumb are about as good as it gets in terms of the physicochemical design principles for a drug carrier. But they offer little insight into how a delivery modality is likely to perform in different tissues and in different disease contexts. They provide little predictive power as to whether a vehicle will induce immunogenicity or other toxicities if taken acutely or chronically. Most of all, they leave us in the dark as to how these molecules interact with and influence cells.

It was these aspects that last month's conference sought to explore. How do macromolecules, viruses or bacterial toxins interact with membrane constituents, such as Rab or ceramide domains? Under what conditions does a particular carrier deleteriously affect viability or replicative capacity? To what extent do agents that facilitate entry by means of endosomal rupture also activate danger signals like ubiquitination or induce apoptosis? Do strategies that co-opt natural receptor-mediated endocytosis pathways deleteriously affect those pathways and send ripples through cellular signal transduction networks? Which adaptors, retrieval proteins, coat proteins and SNARE proteins are involved? How do delivery agents affect the sorting machinery?

Of course, for most drug delivery researchers these are all rather esoteric questions. The pragmatic view is to continue to focus on empirical phenotypic screens of thousands of different peptide, peptoid or lipid moieties to find hits that increase cytoplasmic uptake and subsequent lead optimization. What's the point of understanding biological mechanism?

The point is that the above approach has been tried for several decades and has only taken us so far. Empirical screens may work going forward to create an increasing array of souped-up biobetter molecules. But it seems unlikely that hits from these screens are going to unpick the really hard problems in drug research.

To do that, drug delivery needs to become a discipline that embraces the cell biology of macromolecule uptake, particularly in key target tissues like the brain, the lung, the intestine and the tumor microenvironment. And it needs to study more closely how viruses and bacterial toxins exploit host trafficking machinery to gain access to intracellular targets. After all, these agents have spent a few years more than drug developers finding out how to enter cells. A few million years more.