It is over nine years since the publication of our first,¹ and three years since the second,² analysis of the sources of new and approved drugs for the treatment of human diseases, both of which indicated that natural products continued to play a highly significant role in the drug discovery and development process.

That this influence of Nature in one guise or another has continued is shown by inspection of the information given below, where with the advantage of now over 25 years of data, we have been able to refine the system, eliminating a few duplicative entries that crept into the original data sets. In particular, as behooves authors from the National Cancer Institute (NCI), in the specific case of cancer treatments, we have gone back to consult the records of the FDA and added to these, comments from investigators who have informed us over the past two years of compounds that may have been approved in other countries and that were not captured in our earlier searches. These cancer data will be presented as a stand-alone section as well as including the last 25 years of data in the overall discussion.

As we mentioned in our 2003 review,² the development of high-throughput screens based on molecular targets had led to a demand for the generation of large libraries of compounds to satisfy the enormous capacities of these screens. As we mentioned at that time, the shift away from large combinatorial libraries has continued, with the emphasis now being on small, focused (100 to ~3000) collections that contain much of the "structural aspects" of natural products. Various names have been given to this process, including "Diversity Oriented Syntheses",^3-6 but we prefer to simply say "more natural product-like", in terms of their combinations of heteroatoms and significant numbers of chiral centers within a single molecule,⁷ or even "natural product mimics" if they happen to be direct competitive inhibitors of the natural substrate. It should also be pointed out that Lipinski's fifth rule effectively states that the first four rules do not apply to natural products or to any molecule that is recognized by an active transport system when considering "druggable chemical entities".^8-10

Although combinatorial chemistry in one or more of its manifestations has now been used as a discovery source for approximately 70% of the time covered by this review, to date, we can find only one de novo new chemical entity (NCE) reported in the public domain as resulting from this method of chemical discovery and approved for drug use anywhere. This is the antitumor compound known as sorafenib (Nexavar, 1) from Bayer, approved by the FDA in 2005. It was known during development as BAY-43-9006 and is a multikinase inhibitor, targeting several serine/threonine and receptor tyrosine kinases (RAF kinase, VEGFR-2, VEGFR-3, PDGFR-beta, KIT, and FLT-3) and is in multiple clinical trials as both combination and single-agent therapies at the present time, a common practice once approved for one class of cancer treatment.

As mentioned by the authors in prior reviews on this topic and others, the developmental capability of combinatorial chemistry as a means for structural optimization once an active skeleton has been identified is without par. The expected surge in productivity, however, has not materialized; thus, the number of new active substances (NASs), also known as New Chemical Entities (NCEs), which we consider to encompass all molecules, including biologics and vaccines, from our data set hit a 24-year low of 25 in 2004 (though 28% of these were assigned to the ND category), with a rebound to 54 in 2005, with 24% being N or ND and 37% being biologics (B) or vaccines (V). Fortunately, however, research being conducted by groups such as Danishefsky's, Ganesan's, Nicolaou's, Porco's, Quinn's, Schreiber's, Shair's, Waldmann's, and Wipf's is continuing the modification of active natural product skeletons as leads to novel agents, so in due course, the numbers of materials developed by linking Mother Nature to combinatorial synthetic techniques should increase. This aspect, plus the potential contributions from the utilization of genetic analyses of microbes, will be discussed at the end of this review.

Against this backdrop, we now present an updated analysis of the role of natural products in the drug discovery and development process, dating from 01/1981 through 06/2006. As in our earlier analyses,^1,2 we have consulted the Annual Reports of Medicinal Chemistry, in this case from 1984 to 2005,^11-32 and have produced a more comprehensive coverage of the 1981-2006 time frame through addition of data from the publication Drug News and Perspective^33-49 and searches of the Prous Integrity database, as well as by including information from individual investigators. We also updated the biologicals section of the data set using information culled from disparate sources that culminated in a recent review (2005) on biopharmaceutical drugs.⁵⁰

We have also included relevant references in a condensed form in Tables 1-5, 8, and 9; otherwise the numbers of references cited in the review would become overwhelming. In these cases, "ARMC ##" refers to the volume of Annual Reports in Medicinal Chemistry together with the page on which the structure(s) can be found. Similarly, "DNP ##" refers to the volume of Drug News and Perspective and the corresponding page(s), and an "I ######" is the accession number in the Prous Integrity database. Finally, we have used "Boyd" to refer to a review article⁵¹ on clinical antitumor agents and "M'dale" to refer to Martindale⁵² with the relevant page noted.

It should be noted that the "Year" header in all tables is the "Year of Introduction" of the drug. In some cases there are discrepancies between sources as to the actual year due to differences in definitions. We have generally taken the earliest year in the absence of further information.

Results

As before, we have covered only New Chemical Entities (NCEs) in the present analysis. If one reads the FDA and PhRMA Web sites, the numbers of NDA approvals are in the high tens to low hundred numbers for the last few years. If, however, one removes combinations of older drugs and old drugs with new indications, and/or improved delivery systems, then the number of true NCEs is only in the 20s to just over 50 per year for the last five or so years (see Figures 2 and 5).

	Figure 1 All new chemical entities, 01/1981-06/2006, by source (N = 1184).
	Figure 2 All new chemical entities organized by source/year (N = 1184).
	Figure 3 All small molecule new chemical entities, 01/1981-06/2006, by Source (N = 974).
	Figure 4 Small molecule new chemical entities organized by source/year (N = 974).
	Figure 5 All available anticancer drugs, 1940s-06/2006, by source (N = 175).

As in our earlier analyses,^1,2 the data have been analyzed in terms of numbers and classified according to their origin using both the previous major categories and their subdivisions.

Major Categories of Sources. The major categories used are as follows:

"B" Biological; usually a large (>45 residues) peptide or protein either isolated from an organism/cell line or produced by biotechnological means in a surrogate host.

"N" Natural product.

"ND" Derived from a natural product and is usually a semisynthetic modification.

"S" Totally synthetic drug, often found by random screening/modification of an existing agent.

"S*" Made by total synthesis, but the pharmacophore is/was from a natural product.

"V" Vaccine.

Subcategory. "NM" Natural product mimic (see rationale and examples below).

(For amplification as to the rationales used for categorizing using the above subdivisions, the reader should consult the earlier reviews.^1,2)

In the field of anticancer therapy, the advent in 2001 of Gleevec, a protein tyrosine kinase inhibitor, was justly heralded as a breakthrough in the treatment of leukemia. This compound was classified as an "NM" on the basis of its competitive displacement of the natural substrate ATP, whose intracellular concentrations can approach 5 mM. We have continued to classify PTK and other kinase inhibitors that are approved as drugs under the "/NM" category for exactly the same reasons as elaborated in the 2003 review² and have continued to extend it to cover other direct inhibitors/antagonists of the natural substrate/receptor interaction whether obtained by direct experiment or by in silico studies followed by direct assay in the relevant system. Similarly, a number of new peptidic drug entities, though formally synthetic in nature, are simply produced by synthetic methods rather than by the use of fermentation or extraction. In some cases, an end group might have been changed for ease of recovery. In addition, a number of compounds produced totally by synthesis are, in fact, isosteres of the peptidic substrate and are thus "natural product mimics" in the truest sense of the term. For further information on this area, interested readers should consult the excellent review by Hruby.⁵³

As an example of what can be found by studies around relatively simple peptidomimics of the angiotensin II structure, the recent paper of Wan et al.⁵⁴ demonstrating the modification of the known but nonselective AT₁/AT₂ agonist L-162313 (2, itself related to the sartans) into the highly selective AT₂ agonist (3) (a peptidomimetic structure) led to the very recent identification of short pseudopeptides exemplified by 4, which is equipotent (binding affinity = 500 pM) with angiotensin II and has a better than 20 000-fold selectivity versus AT₁, whereas angiotensin II has only a 5-fold binding selectivity in the same assay.⁵⁵ It will be interesting to see if any compounds such as these will end up as cardiovascular agents since it has been demonstrated that activation of the AT₂ receptor affects cardiac remodeling and leads to reduced blood pressure.⁵⁶

In the area of modifications of natural products by combinatorial methods to produce entirely different compounds that may bear little if any resemblance to the original, but are legitimately assignable to the "NM" category, citations are given in previous reviews.^3,57-64 In addition, one should consult the recent reports from Waldmann's group^65,66 and those by Ganesan,⁶⁷ Shang and Tan,⁶⁸ Constantino,⁶⁹ and Violette et al.⁷⁰ on the use of privileged structures as skeletons around which to build libraries. Another paper of interest in this regard is the editorial by Macarron from GSK,⁹ as this may be the first time where data from industry on the results of HTS screens of combichem libraries versus potential targets were reported with a discussion of lead discovery rates. In this paper, Macarron reemphasizes the fifth Lipinski rule, which is often ignored; "natural products do not obey the other four".

Overview of Results. The data that we have analyzed in a variety of ways are presented as a series of bar graphs and pie charts and two major tables in order to establish the overall pictures and then are further subdivided into some major therapeutic areas using a tabular format. Except where noted, the time frame covered was 01/1981-06/2006:

New Approved Drugs: With all source categories (Figure 1)

New Approved Drugs: By source/year (Figure 2)

Sources of all NCEs: Where four or more drugs were approved per medical indication (Table 1)

Sources of Small Molecule NCEs: All subdivisions (Figure 3)

Sources of Small Molecule NCEs: By source/year (Figure 4)

Antibacterial Drugs: Generic and trade names, year, reference, and source (Table 2)

Antifungal Drugs: Generic and trade names, year, reference, and source (Table 3)

Antiviral Drugs: Generic and trade names, year, reference, and source (Table 4)

Antiparasitic Drugs: Generic and trade names, year, reference, and source (Table 5)

Antiinfective Drugs: All molecules, source, and numbers (Table 6)

Antiinfective Drugs: Small molecules, source, and numbers (Table 7)

Anticancer Drugs: Generic and trade names, year, reference, and source (Table 8)

All Anticancer Drugs: Generic names, reference, and source (Figures 5-7; and (1940s-06/2006) Table 9)

	Figure 6 Approved anticancer agents, organized by source/year (known dates for 157).
	Figure 7 Approved anticancer agents, organized by source/year (unknown dates for 18).

Antidiabetic Drugs: Generic and trade names, year, reference, and source (Table 10)

The extensive data sets shown in the figures and tables referred to above highlight the continuing role that natural products and structures derived from or related to natural products from all sources have played and continue to play in the development of the current therapeutic armamentarium of the physician. Inspection of the data shows this continued important role for natural products in spite of the current low level of natural products-based drug discovery programs in major pharmaceutical houses.

Inspection of the rate of NCE approvals as shown in Figure 2 demonstrates that the natural products field is still producing or is involved in ~50% of all small molecules in the years 2000-2006 and that a significant number of NCEs are biologicals or vaccines (83 of 264, or 31.4%). This is so in spite of many years of work by the pharmaceutical industry devoted to high-throughput screening of predominately combinatorial chemistry products and that the time period chosen should have provided a sufficient time span for combinatorial chemistry work from the late 1980s onward to have produced approved NCEs.

Overall, of the 1184 NCEs covering all diseases/countries/sources in the years 01/1981-06/2006, and using the "NM" classifications introduced in our 2003 review,^1,2 30% were synthetic in origin, thus demonstrating the influence of "other than formal synthetics" on drug discovery and approval (Figure 1).

Inspection of Table 1 demonstrates that, overall, the major disease areas that have been investigated (in terms of numbers of drugs approved) in the pharmaceutical industry continue to be infectious diseases (microbial, parasitic, and viral), cancer, antihypertensives, and antiinflammatory indications, all with over 50 approved drug therapies. It should be noted, however, that numbers of approved drugs/disease do not correlate with the "value" as measured by sales, since the best selling drug of all is atorvastin, a hypocholesterolemic descended directly from a natural product, which sold over $11 billion in 2004 and is at or above this level even today.

The major category by far is that of antiinfectives including antiviral vaccines, with 230 (22.8%) of the total (1010 for indications 4) falling into this one major human disease area. On further analyses (Tables 6 and 7), the influence of biologicals and vaccines in this disease complex is such that only a little over 30% are synthetic in origin. If one considers only small molecules (reducing the total by 50 to 180; Table 10), then the synthetic figure goes up to 31.1%, marginally greater than in our previous report.² As reported previously,^1,2 these synthetic drugs actually tend to be of two basic chemotypes, the azole-based antifungals and the quinolone-based antibacterials.

Four small molecule drugs were approved in the antibacterial area from 01/2003 to 06/2006. These included daptomycin (N, 5) from Cubist, a lipopeptide whose biosynthetic cluster has been successfully cloned and expressed by investigators associated with Cubist.⁷¹ Wyeth had their modified tetracycline derivative, tigecycline, approved (ND, 6), a drug designed to overcome the tet resistance pump in pathogenic bacteria, and another carbapenem (ND) and a quinolone (S) were also approved in this time frame. In the antifungal area, of the five drugs approved, four were azoles (S) and the echinocandin derivative, anidulofungin (ND), was approved for use in the U.S. in early 2006. In the antiviral area, seven drugs were approved for HIV treatment (1 ND, 1 S*, 5 S*/NM). It is interesting that the one ND, enfuvirtide, though listed in most literature as a synthetic, is actually the "end-capped" 36-residue peptide that corresponds to residues 643-678 of the HIV-1 transmembrane protein gp41 and blocks viral fusion with the cell.⁷² In addition to this novel mechanism, four new HIV protease inhibitors were approved; all were peptidomimetics imitating the peptide substrate, and the latest one, darunavir (7), actually has the hydroxyethyl isostere that was first identified in the microbial aspartic protease inhibitor pepstatin and incorporated in the base structure of crixivan (see discussion by Yang et al.⁷³).

It should be noted that the percentages used in the following overall analyses do not always agree with those in the later tables, as all sources, which include B and V categorized drugs, and all indications are included in the percentage figures used in the analyses. Much fuller details are given in the Supporting Information in the form of an Excel XP spreadsheet.

As we reported in our earlier analyses,^1,2 there are still significant therapeutic classes where the available drugs are totally synthetic at the present time. These include antihistamines, diuretics, and hypnotics for indications with four or more approved drugs (cf. Table 1). There are a substantial number of indications where there are three or less drugs that are also totally synthetic. Because of our introduction of the "NM" subcategory, indications such as antidepressants and cardiotonics now have substantial numbers that, although formally "S", now fall into the "S/NM" subcategory.

From inspection of Tables 1-4 and 8 and the Excel XP spreadsheet, the following points can be made in addition to the digest on antiinfectives given in Tables 6 and 7. In the antibacterial area (Table 2), as found previously, the vast majority of the 98 small molecule NCEs are N (10; 10.2%), ND (64; 65.3%), or S*/NM (1; 1%), amounting to 75 in total, or 76.5% of the whole, with the remainder (S) being predominately quinolones. In the antifungal area (Table 3), the roles of the small molecules (n = 28) are reversed, with the great majority being S (22; 78.6%) and S/NM (3; 10.7%), with the remainder being ND (3; 10.7%).

In the antiviral area (Table 4), the situation is somewhat different, with a large number of vaccines (n = 25) now added to this category. If we consider only small molecules, the anti-HIV drugs being approved are based mainly on nucleoside structures (S*) or on peptidomimetics (S* and S/NM), and drugs against other viral diseases also fall into these categories. Thus, one can see that of the 42 small molecule approved antiviral agents, the relevant figures are ND (2; 4.8%), S* and S*/NM categories (32; 76.2%), with the remainder falling into either S (7; 16.7%) or S/NM (1; 2.4%).

We have also identified the antiparasitic drugs that have been approved over the years (Table 5) and point out that of the 14 small molecule drugs, only four are synthetic (28.5%) and of the rest, three are artemisinin derivatives. What is of interest with this base structure is that, in addition to their known antimalarial activities, compounds based on this structure are demonstrating activity as antitumor agents.⁷⁴

With anticancer drugs (Table 8), where in the time frame covered (01/1981-06/2006) there were 100 NCEs in toto, the number of nonbiologicals was 81 (81%). These small molecules could be divided as follows (using 81 = 100%) into N (9; 11.1%), ND (25; 30.9%), S (18; 22.2%), S/NM (12; 14.8%), S* (11; 13.6%), and S*/NM (6; 7.4%). Thus, using our criteria, only 22.2% of the total number of anticancer drugs were classifiable into the S (synthetic) category. Expressed as a proportion of the nonbiologicals/vaccines, then 63 of 81 (77.8%) were either natural products per se or were based thereon, or mimicked natural products in one form or another.

In this current review, we have continued as in our previous contribution (2003)² to reassess the influence of natural products and their mimics as leads to anticancer drugs. By using data from the FDA listings of antitumor drugs, coupled with our previous data sources and with help from Japanese colleagues, we have been able to identify the years in which all but 18 of the 175 drugs we have listed in Table 9 were approved. We have identified these other 18 agents by inspection of three time-relevant textbooks on antitumor treatment,^51,75,76 and these were added to the overall listings using the lead authors' names as the source citation.

Inspection of Figures 5-7 and Table 9 shows that, over the whole category of anticancer drugs effectively available to the West and Japan, the 175 available agents can be categorized as follows: B (18; 10%), N (25; 14%), ND (48; 28%), S (42; 24%), S/NM (14; 8%), S* (20; 11%), S*/NM (6; 4%), and V (2; 1%). If one removes the biologicals and vaccines, reducing the overall number to 155 (100%), the number of naturally inspired agents (i.e., N, ND, S/NM, S*, S*/NM) is 113 (72.9%). It should be noted that these 155 agents do not include some of the earlier drugs that were really immuno- or hematologic stimulants. Etoposide phosphate is not included in this count, as it is a prodrug of etoposide, though it was included in our last review as an approved NCE. We have however included paclitaxel nanoparticles, as this is not just a salt form but is a novel form of the agent ensuring much better water solubility.

In our earlier papers, the number of nonsynthetic antitumor agents was 62% for other than biologicals/vaccines, without an "NM" subcategory. The corresponding figure obtained by removing the NM subcategory in this analysis is 64%. Thus, the proportion has remained similar in spite of some reassignments of sources and the expansion of combinatorial chemistry techniques. As mentioned earlier, the first and only de novo combinatorial drug that we have been able to identify was approved by the FDA in 2005 under the generic name of sorafenib mesylate (1) for the treatment of advanced renal cancer.

A major general class of drugs that was not commented on in any detail in our earlier papers is the class that is directed toward the treatment of diabetes, both types I and II (Table 10; n = 32). These drugs include a significant number of biologics based upon varying modifications of insulin produced in general by biotechnological means (B, 18; 56.3%).⁵⁰ In addition to these well-known agents, the class also includes a very interesting compound (approved by the FDA in 2005) that is assigned to the ND class (extenatide or Byetta). This is the first in a new class of therapeutic agents known as incretin mimetics. The drug exhibits glucose-lowering activity similar to the naturally occurring incretin hormone glucagon-like peptide-1 (GLP-1), but is a 39-residue peptide based upon one of the peptide venoms of the Gila monster, Heloderma suspectum.⁷⁷

Discussion

As alluded to in our previous review, the decline or leveling of the output of the R&D programs of the pharmaceutical companies has continued, with the number of drugs of all types dropping in 2003 to 35 launches, including 13 in the B/V categories, and reaching a nadir in 2004, when only 25 were launches and 6 of these fell into the B/V categories. There was a significant upswing in 2005 with 54 launches, but 20 of these were in the B/V categories, leaving 34 small molecules. In the first 6 months of 2006, of the 22 launches, 9 were B/V.

Although combinatorial chemistry continues to play a major role in the drug development process, as mentioned earlier, it is noteworthy that the trend toward the synthesis of complex natural product-like libraries has continued. As was eloquently stated by Danishefsky in 2002, "a small collection of smart compounds may be more valuable than a much larger hodgepodge collection mindlessly assembled".⁷⁸ Recently he and a coauthor restated this theme:⁷⁹

In summary, we have presented several happy experiences in the course of our program directed toward bringing to bear nature's treasures of small molecule natural products on the momentous challenge of human neurodegenerative diseases. While biological results are now being accumulated for systematic disclosure, it is already clear that there is considerable potential in compounds obtained through plowing in the landscape of natural products. Particularly impressive are those compounds that are obtained through diverted total synthesis, i.e., through methodology, which was redirected from the original (and realized) goal of total synthesis, to encompass otherwise unavailable congeners. We are confident that the program will lead, minimally, to compounds that are deserving of serious preclinical follow-up. At the broader level, we note that this program will confirm once again (if further confirmation is, indeed, necessary) the extraordinary advantages of small molecule natural products as sources of agents, which interject themselves in a helpful way in various physiological processes.

We close with the hope and expectation that enterprising and hearty organic chemists will not pass up the unique head start that natural products provide in the quest for new agents and new directions in medicinal discovery. We would chance to predict that even as the currently fashionable "telephone directory" mode of research is subjected to much overdue scrutiny and performance-based assessment, organic chemists in concert with biologists and even clinicians will be enjoying as well as exploiting the rich troves provided by nature's small molecules.

A rapid analysis of the entities approved from 2003 to 2006 (the full data set is available as an Excel spreadsheet in the Supporting Information) indicated that there were significant numbers of antitumor, antibacterial, and antifungal agents approved as mentioned above. This time frame also saw two very important approvals, both of which were natural products. The first was the approval by the FDA, after a long series of trials and discussions, of the cone snail toxin known as Prialt, which is the first "direct from the sea" entity to become a licensed pharmaceutical.^80,81 Although one can argue (as we have on other occasions) that the discovery of the arabinose nucleosides by Bergmann in the 1950s was the driving force behind Ara-A, Ara-C, AZT, etc., this is the first direct transition from marine invertebrate to man. Also in the middle of 2006, the botanical preparation Hemoxin^82,83 was approved in Nigeria following demonstration of efficacy in clinical trials as a treatment for sickle cell anemia. This is a mix of plants that came from native healer information and thus can be classified as a "true ethnobotanical preparation".

In this paper, as we stated in 2003,² we have again demonstrated that natural products play a dominant role in the discovery of leads for the development of drugs for the treatment of human diseases. Some have argued (though not in press, only in personal conversations at various fora) that the introduction of categories such as S/NM and S*/NM is an overstatement of the role played by natural products in the drug discovery process. On the contrary, we would argue that these further serve to illustrate the inspiration provided by Nature to receptive organic chemists in devising ingenious syntheses of structural mimics to compete with Mother Nature's longstanding substrates. Even discounting these categories, the continuing and overwhelming contribution of natural products to the expansion of the chemotherapeutic armamentarium is clearly evident, and as we stated in our earlier papers, much of Nature's "treasure trove of small molecules" remains to be explored, particularly from the marine and microbial environments.

From the perspective of microbes and their role(s) as sources of novel bioactive entities, the recent work that has been reported by a variety of investigators as to the potential of these organisms needs to be widely disseminated. Over the last few years, it has become obvious from analyses of the published (and, to some extent, unpublished) genomic sequences of a variety of microbes that there are at least a dozen potential biosynthetic clusters in each organism surveyed and, in certain well-publicized cases, over 30 such groupings.^84-92 In the marine environment the interplay of these two sources, as exemplified by the recent review by Newman and Hill,⁹³ leaves no doubt that a host of novel, bioactive chemotypes await discovery from both terrestrial and marine sources.

In this respect it should be noted that in the last year or so there has been a very significant series of findings where the well-known antitumor agents camptothecin⁹⁴ and podophyllotoxin⁹⁵ and vincristine⁹⁶ have now been produced by fermentation of endophytic fungi, isolated from the producing plants. The usual argument that these are artifacts because of the inability to produce large quantities by regular fermentation processes has been shown to be specious by the work by Bok et al.⁸⁴ with Aspergillus nidulans. This work demonstrated that one has to be able to find the "genetic on switch" to be able to obtain expression of such clusters outside of the host. In addition to these papers the reader's attention is also drawn to the recent excellent review article by Gunatilaka⁹⁷ on this subject, which gives an excellent overview of the numbers of materials so far discovered from these sources. As a result, investigators need to consider all possible routes to novel agents.

To us, a multidisciplinary approach to drug discovery, involving the generation of truly novel molecular diversity from natural product sources, combined with total and combinatorial synthetic methodologies, and including the manipulation of biosynthetic pathways (so-called combinatorial biosynthesis), provides the best solution to the current productivity crisis facing the scientific community engaged in drug discovery and development.

Once more, as we stated in our 2003 review,² we strongly advocate expanding, not decreasing, the exploration of Nature as a source of novel active agents that may serve as the leads and scaffolds for elaboration into desperately needed efficacious drugs for a multitude of disease indications.