If liposomes are to realize their potential as drug-delivery vehicles, techniques must be developed to target them to particular cell types or anatomical locations. The usual strategy for targeting liposomes is to attach a ligand specific for a determinant on the target cell. Antigens, immunoglobulins, glycoproteins, glycolipids, and lectins have all been used in this way. 1" 10 Alternative methods ofliposome targeting use the inherent tendency of liposomes to be taken up by phagocytes 11 or localized in specific anatomical compartments (e.g., lung, joints, peritoneal cavity, lymphatics). 12 Each of these approaches has its limitations. In ligand-mediated targeting, specific binding of liposomes to the cell does not guarantee that the contents of the liposome will reach the cytoplasm. 6 In addition, the presence of endothelia or other histological barriers between liposome and binding site may restrict this method to applications in which the liposomes are administered directly into the target compartment. Immunopotentiating agents have been delivered selectively to phagocytes in the lung, 13 significantly retarding the development of pulmonary metastasis. 14 However, the majority of the liposomes are still taken up by the liver and spleen. Compartmental targeting has shown promise for delivering steroids to arthritic joints, 15 and anticancer drugs to peritoneal and lymphatic sites. 16

In collaboration with Yatvin and other co-workers we have explored a conceptually very different approach to "targeting". The idea is not to obtain specific localization of the liposomes themselves but rather to have them release their contents in a specified capillary bed, for example, that of a tumor. Liposomes are known to release encapsulated watersoluble contents more quickly near their liquid crystalline phase-transition temperature (T m) than at other temperatures. 17 . 21 Selective release can therefore be obtained in a locally heated region by injecting liposomes designed to have Tm a few degrees above physiological temperatures (see Figure 1). For purposes of identification in the pharmacological context we term such liposomes ''temperature sensitive''. We have described elsewhere the principles of their in vivo use in conjunction with local hyperthermia. 22 ·23

Temperature-sensitive release can be engineered by the selection of pure lipids that undergo sharp transition temperatures or by using mutually miscible mixtures of pure lipids to adjust the transition temperature to the desired point. For these studies we have used dipalmitoyl phosphatidylcholine (DPPC) (T m = 41 °C), usually in combination with dipalmitoyl phosphatidylglycerol (DPPG) (Tm = 41°C), or distearoyl phosphatidylcholine (DSPC) (Tm = 54 °C). The choice of lipids and the relative proportion of each depend upon the desired T m and the size of the liposome. Small unilamellar vesicles (SUV) have apparent transition temperatures several degrees below those predicted from the T m of the component lipids in large unilamellar vesicles (LUV) or in multilamellar vesicles (ML V). This effect is probably a result of stress in the highly curved bilayer structure. 24

Although each of the principle types of liposome (ML V, LUV, SUV) exhibits temperaturesensitive drug release given an appropriate choice of lipids, the process is not exactly the same for all three: release characteristics such as the sharpness of the transition, the rate of drug release, and the influence of serum components differ. These factors, which will be discussed here in some detail, must be considered when designing temperature-sensitive liposomes for a particular application.