Article Figures & Data
Figures
- Fig. 1
Clinical trials and drugs on brain cancer. (A) Number of unique agents entering clinical trials per year for treatment of brain cancer. Each bar corresponds to the number of unique agents that appeared for the first time each year from 2010 to 2020. The peak of 2010 could be partially explained by the fact that some of the agents had already entered trials before 2010 and are included in the year 2010, as they were still identified in trials when we began our analysis. The last year contains a small number of agents since it has not been completed at the time of writing. Data are to February 2020. (B) The number of clinical trials registered per year. Each bar corresponds to the number of clinical trials registered in ClinicalTrials.gov per year. The color indicates the phase of a trial. Note, phases were treated as mentioned in the section on data collection and analysis. Data are to February 2020.
- Fig. 2
Structural platforms in brain cancer trials. The structural classification of the drug agents. The structural classification of the 568 drug entities identified in the analysis are separated into six superclasses. Some of the entities represent named combinations, thus their structure is shown using “&.” The size of a bar corresponds to the number of agents in the class. Each classification has color codes to illustrate the superclasses of entities. The right side of the figure details the diversity of delivery systems for small molecules in the data. Data are to February 2020. BITE, bispecific T-cell engager; CP, cell product; Cyt., cytotoxic; DS, delivery system; NKT cell, natural killer T cell; siRNA, small interfering RNA.
- Fig. 3
Key structural platforms. Key structural platforms and classes of agents. This figure provides a graphical representation of the key structural classes identified in clinical trials on primary brain malignancies. Small-molecule platforms may contain delivery systems that could be of different origins. Antibody-based platforms contain bispecific T-cell engager therapies wherein the domains bind with different antigens. In several cell-based platforms, the key cell-surface markers that are used to sort these cell types were illustrated. T cells could possess different versions of a T-cell receptor—αβ or γδ, depending on a subtype. TCR T cells use an artificial T-cell receptor that is typically composed of a single protein chain. The TCR of the NKT cells is considered to be CD1d complementary and response to lipid antigens. The 2nd generation CAR structure for CAR NK and CAR T cells was used since it is the most common form of this construct. Dendritic cells are used as a vaccine after pulsed with tumor antigens. Bacterial cells are used to express tumor peptides and deliver them to immune cells to trigger the antitumor immune response. Viral-based platforms include different types of viruses that can be used in different strategies, such as adenovirus and HSV that are both used as oncolytic viruses or as a vector for gene therapy. Their genomes could be genetically edited to insert a gene of interest that could activate the immune response, trigger the apoptosis of tumor cells, or confer sensitivity to chemotherapy. Viruses are illustrated schematically and may not represent the actual structure in detail. FAS-TNFR1, chimeric FAS and tumor necrosis factor receptor 1; NKT cell, natural killer T cell; siRNA, small interfering RNA; TK, thymidine kinase.
- Fig. 4
Pharmacological directions of therapies. The pharmacological classification of the drug agents. The 568 drug entities identified in the analysis are classified by pharmacological classes into five superclasses, which are color-coded. Information was obtained using cancer.gov, fda.gov, and PubMed search. Note that if a kinase inhibitor exhibits very clear antiangiogenic activity, it is classified as angiogenesis inhibitors. The right side of the figure provides secondary classification for other therapies (that has not been used for trend analysis). Data are to February 2020. By the term “genetically unmodified effector cells” we mean cells with unmodified immune effector function. BITE, bispecific T-cell engager; CBT, checkpoint blockade therapy; CT, chemotherapy; CXCR4, C-X-C chemokine receptor type 4; ER, expression regulator; Gen, genetically; GT, gene therapy; IT, immune therapy; NMDA, N-methyl-d-aspartate receptor; NSAID, nonsteroidal anti-inflammatory drug; Synth., synthetic.
- Fig. 5
Targets and their distribution. (A) The distribution of target locations. Each target location was manually obtained from the UniProt database and PubMed search. If a target had multiple cellular localizations, we included the most relevant for its functional activity. If the exact function is unknown or hard to determine, we used the location with the largest number of references. Color corresponds to a target location. (B) The general functional activities of the drug targets. Each drug target was manually classified using the UniProt database and PubMed searches. Targets are color-coded corresponding to seven major functional classes. (C) Specific descriptions of each class of drug target presented in part B. Transporters were classified using the Transporter Classification Database. The term “kinase” refers to all kinases, excluding RTK and nontyrosine receptor kinases. Struct., structural; Synth., synthesis.
- Fig. 6
Drug-target network. The drug-target network and interactions. The area in the center with magenta edges is the giant component. The giant component is divided into several superclusters. Colors of the nodes indicate the following: red is targeted therapy, green is chemotherapy, blue is immune therapy, yellow is gene therapy, gray is other therapies, and vinous is target. This network has been created via R programming language. Data are to February 2020.
- Fig. 7
Combination analysis. (A) Number of unique combinations per therapeutic class. Each bar corresponds to the median (green color) or the mean number (red color) of unique combinations for each drug in a particular therapeutic class. Bold numbers in parentheses are the number of agents in each corresponding class. The large difference among means and medians indicates that combinational designs are skewed toward particular drugs. For instance, in alkylating agents, this difference is related to the fact that temozolomide has been detected in 209 unique combinations (more than 30% of all designs). (B) Schematic representation of network algorithm. The schematic representation of the algorithm forming a combination network from the input vector of all combinations. Each node corresponds to the unique combination. If one combination is a subcombination of another combination, the algorithm creates an edge among two nodes. (C) The percentage of combinational and monotherapy arms among clinical trials of different phases. The number below each pie chart shows the number of treatment arms. This analysis is quantitative and does not remove the same combination designs from different clinical trials. By the term “genetically unmodified effector cells” we mean cells with unmodified immune effector function. BITE, bispecific T-cell engager; CBT, checkpoint blockade therapy; CT, chemotherapy; ER, expression regulators; Gen, genetically; GT, gene therapy; IT, immune therapy; Synth., synthetic.
- Fig. 8
Quantitative trends for phase one clinical trials. Trends in phase one clinical trials. This figure illustrates the trends for the most significant classes in phase one clinical trials of the dataset. The presence was calculated using the combined number of all drug entities of a particular class per year. If a particular drug appeared in a clinical trial, it was counted only once, no matter the number of arms with this agent in a trial. If a particular drug was analyzed in several different clinical trials, each of the trials was counted. If several agents of the same class appeared in the same trial, it is counted as the number of these agents. The colors of trend lines correspond to pharmacological classes. Data are to December 2019. CBT, checkpoint blockade therapy; CT, chemotherapy; ER, expression regulator; GT, gene therapy; IT, immune therapy.
Tables
- TABLE 1
Selected key structural classes
This table describes the key structural platforms used in brain cancer clinical trials and their core advantages and limitations.
Superclass Subclasses Advantages Limitations Useful Resources Small molecules Small molecules
Small molecules + delivery system
Low molecular weight is favorable for penetrating the BBB
Many available screening approaches
Easy to use drug formulations can be available
Targeting is not limited to membrane targets
On-target and off-target toxicities
Clearance might be rapid, requiring an additional administration of an agent
The delivery system might possess additional toxicities
(Ferguson and Gray, 2018; Macarron et al., 2011; Ovacik and Lin, 2018; Scherrmann, 2002) Antibody products and derivatives Whole antibody
Phage-display libraries allow the relatively easy design of a wide spectrum of antibodies (all antibody products)
Long blood circulation, hence administration is less frequent
Tumor antigen escape (all antibody products)
Distribution is typically limited to blood and interstitial fluids
The target spectrum is usually limited to the membrane and free-floating proteins (all antibody products)
On-target off-tissue toxicity (all antibody products)
Immunogenicity that might lead to anaphylaxis and cytokine release syndrome (CRS) (all antibody products)
(Ovacik and Lin, 2018) Antibody-drug conjugate
Combine the potency of small molecules with the selectivity of antibodies
Reduced off-target toxicity of the cytotoxic molecule
Some versions possess bystander killing activity
Linker design might help to overcome drug resistance
Deconjugation could cause toxicities
The quantity of cytotoxic agent delivered is typically low
(Beck et al., 2017) Radio-antibody conjugate
Therapeutic and diagnostic tool
The radioactive exposure is limited to a tumor site
Tumor exposure can be low
The amount of exposure is not easy to quantify
Radionuclides could be uptaken by a nontarget tissue/organ
(Bourgeois et al., 2017; Vivier et al., 2018) Antibody fusion product
Depending on the attached component this molecule could act as a vaccine, targeted toxin, or immune modulator
Molecular weight is higher than those of antibodies, and hence tissue penetration could be challenging
(Balza et al., 2010; Bao et al., 2016; Dhodapkar et al., 2014) Bispecific T-cell engager (BiTE)
Tumor antigen does not require MHC presentation to elicit the immune response
Lower molecular weight favors tumor penetration
Short-term serum half-life that requires relatively frequent administration of an agent
Immunogenicity that might lead to anaphylaxis and cytokine release syndrome
(Baeuerle and Reinhardt, 2009; Labrijn et al., 2019) Peptibody
Peptide could possess various pharmacological functions
Fc domain improves the pharmacokinetic properties of a peptide and provides immune effector function
Immunogenicity is higher compared with peptides and may be undesired
(Cavaco et al., 2017; Shimamoto et al., 2012) Protein and peptide molecules Peptide
Peptide antigens are easier to produce than full proteins
Peptide vaccines induce better responses than full protein vaccines
Pharmacokinetic properties are mediocre; renal excretion is an issue
Stability is limited
The use of adjuvants is required for proper immunogenicity (for vaccines)
(Cavaco et al., 2017; Hos et al., 2018; Mahmood and Green, 2005) Recombinant human protein
Immunogenicity is lower compared with nonhuman proteins
Pharmacokinetic properties may be an issue, thus modifications may be required
(Mahmood and Green, 2005) Protein fusion product
If a protein is conjugated with an Fc domain of an antibody, its pharmacokinetic properties (half-life) are improved
One of the protein components could be used as a targeting moiety to deliver a toxin
Molecular weight can be high, hence cell permeability is limited
Immunogenicity may be an issue
(Kawakami et al., 2003; Sperinde et al., 2020; Strohl, 2015) Protein-drug conjugate
Protein component is used for targeted delivery of a small molecule
Conjugation may help delivery through BBB
Similar to antibody-drug conjugates
(Thomas et al., 2009; Venepalli et al., 2019; Vhora et al., 2015) Cell products Dendritic cells
Multiple antigen coverage
The immune response is generally stronger compared with peptide vaccines
Formulations could be personalized
Different options to load tumor antigens are available
Antigens are MHC-restricted
Production is complex, expensive, and hard to automate and unify
Tumor microenvironment still could diminish the efficacy
(Sabado et al., 2017) CAR T cells
Complete response rates in B cell malignancies were as high as 90%
Primary brain tumors may be a promising area for CAR T therapy given a relatively low mutation burden of these tumors
Targets are not MHC-restricted
Phage-display libraries allow the relatively easy design of a wide spectrum of recognition domains for CAR T cells
Recognition domains could be also composed using other molecules than ScFv
Production is complex, expensive, and hard to automate and unify
Tumor antigen escape renders CAR T cells completely ineffective if occurs
Severe and even lethal systemic toxicities have been observed
On-target off-tumor toxicity
Lack of efficacy in solid tumors
Targeting is limited to membrane tumor antigens
(Aijaz et al., 2018; Chandran and Klebanoff, 2019; Jackson et al., 2016; Levine et al., 2016; Rafiq et al., 2020; Sadelain et al., 2017) Stem cells
Many possible applications, which include drug delivery, chemoprotection, hematopoiesis, chemosensitization, immune modulation
Some cell subsets can possess tumor tropism
Limited availability
Production is complex
Invasive administration
Safety may be an issue
(Aijaz et al., 2018; Bexell et al., 2013; Mount et al., 2015; Parker Kerrigan et al., 2018) T cells
Multiple antigen coverage
Formulations could be personalized
Depending on amplification and modifications these cells could acquire additional properties, such as chemotherapy resistance
Additional stimulation (IL-2) is usually needed
Antigens are MHC-restricted
Production is complex, expensive, and hard to automate and unify
Severe systemic toxicities may occur
T cells may not be isolated from the initial tumor
Tumor microenvironment could diminish the efficacy
NCT04165941, (Met et al., 2019; Stroncek et al., 2019) Modified bacterial cells
Relatively high immunogenicity
Manufacturing is scalable
Toxicities may be high
Risk of undesired infections
Antivector immune response
(Lopes et al., 2019) NK cells
In contrast to T cells, preimmunization is not required
Tumor recognition is different compared with T cells
Immunomodulatory function
Additional stimulation is usually needed
Production is complex, expensive, and hard to automate and unify
Invasive administration
Tumor microenvironment could diminish the efficacy
(Fang et al., 2017; Hodgins et al., 2019) Tumor lysate
Multiple antigen coverage
Formulations could be personalized
Production is relatively simple (compared with other cell-based products)
Immunogenicity could be low
These formulations can lead to an autoimmune response
Allogeneic vaccine sources might have different antigen composition than the patient’s tumor
Tumor may not be surgically available to extract antigens
Lysates may have immune-suppressive molecules from tumor cells
(Gonzalez et al., 2014; Olin et al., 2014; Rojas-Sepulveda et al., 2018) TCR T cells
Compared with CAR T cells, TCR T cells can be targeted against intracellular targets
Less antigen density is required to trigger the immune response
Downstream receptor signaling may be more persistent compared with CAR
Antigens are MHC-restricted
Production is complex, expensive, and hard to automate and unify
Severe and even lethal systemic toxicities have been observed
Tumor microenvironment could diminish the efficacy
(Aijaz et al., 2018; Chandran and Klebanoff, 2019; D'Ippolito et al., 2019) CAR NK cells
Because of the biologic nature of NK cells, the allogeneic application might be safer for CAR NK cells rather than CAR T cells
Primary brain tumors may be a promising area for CAR NK therapy given a relatively low mutation burden of these tumors
Targets are not MHC-restricted
Phage-display libraries allow the relatively easy design of a wide spectrum of recognition domains for CAR NK cells
Recognition domains could be also composed using other molecules than ScFv
Common CAR NK cell line NK-92 is derived from non-Hodgkin lymphoma, thus raises safety concerns
CAR NK cells are more sensitive to cryopreservation than T cells
Other limitations similar to CAR T cells
(Burger et al., 2019; Wang et al., 2020) Bi-armed T cells
Bispecific antibodies could be designed against numerous antigens
Antigens are not MHC-restricted
Severe systemic toxicities have been observed
On-target off-tumor toxicity
Targeting is limited to membrane tumor antigens
(Lum and Thakur, 2011; Zitron et al., 2013) NKT cells
Immunomodulatory function
Tumor recognition is different compared with classic T cells and NK cells
Limited availability of NKT cells in cancer patients
Tumor microenvironment could diminish the efficacy
(Chen et al., 2018; Nair and Dhodapkar, 2017; Terabe and Berzofsky, 2018) Tumor cells
Extracellular vesicles released by a vaccine are a good source of tumor antigens
Formulation is personalized
Immunogenicity of exosomes is considered stronger than those of peptides and lysates
Tumor-derived vesicles may be immunosuppressive
This vaccine is implanted compared with other formulations
(Harshyne et al., 2015; Robbins and Morelli, 2014; Tarasov et al., 2019a) Viral therapeutics Adenovirus
HSV
Measles virus
Parvovirus
Poliovirus
Reovirus
Vaccinia virus
Many viruses can initiate the antitumor immune response
Adenoviral, HSV genomes rarely integrate into the host genome
HSV has many receptors for cell binding and entry
The HSV genome is large and allows to incorporate large genes
The HSV can be effectively controlled by common antiviral drugs
Usually, HSV faster degrades cancer cells than adenoviruses
Measles virus, Parvovirus, and Reovirus possess a natural tumor tropism
Viruses could be targeted at specific cells using engineered proteins and peptides
Lack of efficacy is common
Antivector immune response could limit the viral efficacy
Safety may be an issue
Many patients have antibodies against adenoviral vectors
Adenoviruses are sequestrated by nontarget cells
Manipulations with the genome sizes can decrease the stability of the virus
Adenoviral genome is relatively small (36 kb)
HSV genetic modification is difficult
Some viruses integrate their genome into the host cells
(Bretscher and Marchini, 2019; Foreman et al., 2017; Gromeier and Nair, 2018; Hajeri et al., 2020; Msaouel et al., 2009; Saha et al., 2014; Watanabe and Goshima, 2018) Other DNA plasmid
Multiple antigen coverage
No MHC restriction
Stability is a concern
Relatively weak immunogenicity
The delivery vehicle may be required
(Lopes et al., 2019) Aptamer
Many possible targets
Easy and cheap large-scale synthesis
Specificity
Can be generated to cross the BBB
Limited stability and short half-life
Delivery vehicle is required
Antiformulation immune response
(Cesarini et al., 2020; Zhu and Chen, 2018) - TABLE 2
Trends in most common drug targets
This table illustrates the top 20 targets for drugs other than vaccines and advanced biologicals, like CAR T cells. Note, proteins binding with the Fc region of antibodies have been removed from the table. The presence is calculated by the number of trials in which a particular target has been drugged. If a single trial contains more than one agent that targets a particular drug target, this trial is counted as the number of these agents. Max phase column illustrates the maximal phase of a clinical trial in which a particular target has appeared. Trials with phase four have not been included because of the fact they contain agents that were never approved for brain tumors, and likely their phase is not equal for all analyzed treatment arms. Approved drug targets are labeled as approved. The number of clinical trials is calculated by the number of trials in which a particular target has been drugged, and each trial is counted only once, no matter the number of agents against the target. The number of unique drugs illustrates the number of drugs that exploit the target. For each drug, both pivotal and secondary targets were taken into account. The trend was calculated as follows: if a 3-year avg. of the target presence during 2017, 2018, and 2019 is 10% or more higher than the avg. during the 10-year period, the trend is upward, whereas if it is 10% or more below than the avg. during the 10-year period, the trend is downward. The year column illustrates the first year the target appeared during the analyzed 10 years of brain cancer trials. Data are to February 2020.
Target Presence Max. Phase Number of Trials with Target Number of Unique Agents Drugging the Target Trend Earliest Year DNA 422 Approved 370 27 No trend 2010 VEGFA 109 Approved 108 3 Downward trend 2010 PDCD1 78 3 78 7 Upward trend 2013 NR1I2 64 3 62 14 No trend 2010 KDR 60 2 57 19 No trend 2010 PTGS2 59 3 54 9 No trend 2010 PPARG 52 3 51 6 Upward trend 2010 EGFR 50 2 49 25 No trend 2010 MTOR 48 Approved 47 14 Downward trend 2010 PTGS1 46 3 46 5 Upward trend 2010 TUBB1 46 3 45 16 Downward trend 2010 ACAT1 45 3 45 4 Upward trend 2010 FLT1 45 2 45 13 No trend 2010 KIT 45 2 42 13 Downward trend 2010 PDGFRB 42 2 39 11 Downward trend 2010 FLT4 41 2 41 12 Downward trend 2010 HRH2 40 3 40 2 Upward trend 2010 TOP1 36 2 36 7 Downward trend 2010 PDGFRA 34 2 33 12 Downward trend 2010 TOP1MT 33 2 33 4 Downward trend 2010 TOP2A 32 3 29 11 Downward trend 2010 HDAC2 28 3 28 10 Downward trend 2010 FGFR1 26 2 25 10 Downward trend 2010 TACSTD2 26 2 26 4 Downward trend 2010 TYMS 26 2 23 7 Downward trend 2010 DNA, deoxyribonucleic acid; EGFR, epidermal growth factor receptor; FGFR1, fibroblast Growth Factor Receptor 1; FLT1, vascular endothelial growth factor receptor 1; FLT4, fms-related tyrosine kinase 4; HDAC2, histone Deacetylase 2; KDR, kinase insert domain recepto; KIT, protooncogene c-KIT; MTOR, mechanistic target of rapamycin; NR1I2, nuclear Receptor subfamily 1, group I, member 2; PDCD1, Programmed cell death protein 1; PDGFRA, platelet-derived growth factor receptor A; PDGFRB, plateletderived growth factor receptor beta; TACSTD2, tumor associated calcium signal transducer 2; TOP1, DNA topoisomerase 1; TOP1MT, DNA Topoisomerase I Mitochondrial; TOP2A, DNA Topoisomerase II Alpha; TUBB1, tubulin beta-1 chain; TYMS, thymidylate synthetase; VEGFA, vascular endothelial growth factor A.
- TABLE 3
Promising combination designs
This table illustrates promising combinational designs that have been proposed for selected therapeutic classes and how often these designs are presented in our database. The design pattern column shows a key pattern for a combination, which can also contain additional agents. The number of detected arms shows the number of occurrences for each design pattern among clinical trials. Note, our analysis did not take into account administration schedules and doses. The number for unique designs is written in parentheses. The number of clinical trials column shows the number of clinical trials (by NCTID) for each combination design. The phase structure column and the rational column show the phase counts for each combination design and biologic rationale to use this design respectively. Data are to February 2020.
Design Pattern Number of Detected Arms Number of Clinical Trials Phase Structure Possible Rationale Reference for Rationale Alkylating agents + TTF 12 (7 unique) 12 I – 9 trials
II – 3 trials• Lack of overlapping resistance mechanisms
• Inhibition of DSB-repair mechanism to increase toxicity(Branter et al., 2018) Angiogenesis inhibitors + TTF 6 (3 unique) 6 I – 1 trial
II – 5 trials• Reduced toxicity of angiogenesis inhibitors (Branter et al., 2018) Spindle poisons + TTF Untested Untested NA • Synergetic disruption of spindle functioning (Branter et al., 2018) Immune therapy + TTF 4 (4 unique) 4 I – 1 trial
II – 3 trials• No hindrance on immune response
• Improving infiltration of immune effector cells(Branter et al., 2018) Immune checkpoint inhibitors + DC vaccination 4 (4 unique) 4 I – 2 trials
II – 2 trials• Increased immune response toward neoantigens (Garg et al., 2017; Sabado et al., 2017) Immune checkpoint inhibitors + antibody-drug conjugates Untested Untested NA • Antibody-drug conjugates (their warheads) can induce immunogenic cell death that could be augmented with checkpoint inhibitors
• Combination demonstrates promising results in several tumor types(Beck et al., 2017) Immune checkpoint inhibitors + vaccine (any) 19 (19 unique) 19 I – 11 trials
II – 6 trials• Improved response rates (Barbari et al., 2020) Immune checkpoint inhibitors + radiation therapy 36 (26 unique) 31 I – 19 trials
II – 10 trials
III – 2 trials• Improved response rates
• May overcome resistance
• Antigen release
• Favorable immune-activating environment(Barbari et al., 2020) Immune checkpoint inhibitors + chemotherapy 27 (21 unique) 21 I – 12 trials
II – 9 trials
III – 1 trials• Improved response rates
• Antigen release
• Favorable immune-activating environment(Barbari et al., 2020) Immune checkpoint inhibitors + modified effector cells 4 (4 unique) 3 I – 2 trials
II – 1 trial• Increased persistence of CAR T cells
• Increased antitumor activity(Barbari et al., 2020; Grosser et al., 2019; Jackson et al., 2016) CAR T cells + oncolytic viruses Untested Untested NA • Increased persistence of CAR T cells
• Favorable proinflammatory environment for both CARs and oncolytic viruses(Ajina and Maher, 2019; Guedan and Alemany, 2018; Tang et al., 2020; Twumasi-Boateng et al., 2018) Kinase inhibitor + kinase inhibitor 24 (17 unique) 22 I – 13 trials
II – 7 trials
IV – 2 trials• Tackling with tumor heterogeneity
• Targeting pathway network to overcome drug resistance(Yap et al., 2013) Angiogenesis inhibitors + immune therapy 28 (26 unique; almost all antiangiogenic agents are represented by bevacizumab) 26 I – 12 trials
II – 14 trials• Reduced hypoxia should be favorable for immune response (Ramjiawan et al., 2017) DSB, DNA double-strand breaks; NA, not available.
Data Supplement
- Supplemental Figure 1 -
Supplemental Fig 1 Percentage of discontinued trials. Each bar represents the percent of discontinued clinical trials for each direction. Discontinued trials include Terminated, Suspended, and Withdrawn studies. Since we filtered out all studies with 0 participants, withdrawn studies are not presented in the data. To calculate the overall number of clinical trials for direction, we extracted all clinical trial IDs where members of a drug class have been detected, and counted the number of this vector with clinical trial IDs.
- Supplemental Figure 2 -
Supplemental Fig 2 Target locations. The distribution of target locations. Each target location was manually obtained from the UniProt database and PubMed searches. If a target had multiple cellular localizations, we included the most relevant for its functioning. If the exact function is unknown or hard to determine, we used the location with the largest number of references. Color corresponds to a target location. This figure contains targets for advanced biologicals (cell products, vaccines, and gene therapy) only.
- Supplemental Figure 3 -
Supplemental Fig 3 Combination net. This network visualizes all therapeutic arms that have been detected in the clinical trials. Each node represents a treatment arm, which could be either monotherapy or a combination. Edges are created if one arm is fully presented within another arm. Edges for this network have been calculated via algorithm depicted in figure 7(B). The network has been created using R scripts and the Gephi software.
- Supplemental Figure 4 -
Supplemental Fig 4 Combination net simplified. This network links the agents if they have been found at least in one common combination. Each node represents a treatment or agent. The network has been created using R scripts and the Gephi software.
- Supplemental Figure 5 -
Supplemental Fig 5 Top 50 treatment arm designs. This bar chart shows the presence of the different combination types based on the pharmacological classification of the individual components of a combination. All treatment arms in the data were counted. The color indicates whether the therapeutic strategy is a combination or not. This figure includes only the top 50 therapeutic strategies.
- Supplemental Figure 6 -
Supplemental Fig. 6 Treatment arm designs. This bar chart shows the presence of the different combination types based on the pharmacological classification of the individual components of a combination. All treatment arms in the data were counted. The color indicates whether the therapeutic strategy is a combination or not. This figure is extended and includes all of the therapeutic strategies.
- Supplemental Figure 7 -
Supplemental Fig 7 Exploration of novel targets. This figure illustrates the number of novel targets per year for regular drugs and advanced biologicals (vaccines, cell products, gene therapies). The year 2010 was used as a starting point, so the number of novel targets is technically much higher for this year since it includes targets from previous years. Data is to February, 2020.
- Supplemental Figure 8 -
Supplemental Fig 8 Quantitative trends for drug classes. This figure illustrates the presence of the additional drug classes in phase one clinical trials of the dataset. The presence was calculated using the combined number of all drug entities of a particular class per year. If a particular drug appeared in a clinical trial, it was counted only once, no matter the number of arms with this agent in a trial. If a particular drug was analyzed in several different clinical trials, each of the trials was counted. If several agents of the same class appeared in the same trial, it is counted as the
number of these agents. The colors of trend lines correspond to pharmacological classes. Data is to December, 2019. Abbreviations: PARP, poly (ADP-ribose) polymerase; CBT, checkpoint blockade therapy; IT, immune therapy; GT, gene therapy; ER, expression regulators; CT, chemotherapy; Synth., synthetic. - Supplemental Table 1 -
Supplemental Table 1 Drug classes and features. This table describes the most important features of key drug classes we detected during the analysis of clinical trials on primary brain malignancies.
- Supplemental Table 2 -
Supplemental Table 2 Drug target statistics (regular drugs). This table illustrates the whole list of targets for drugs other than vaccines and advanced biologicals, like CAR T cells. Note, proteins binding with the Fc-region of antibodies have been removed from the table. The presence is calculated by the number of trials, where a particular target has been drugged. If in a single trial contains more than one agent against a particular target, this trial is counted times the number of these agents. Max phase column illustrates the maximal phase of a clinical trial, where a particular target has appeared. Trials with phase IV have not been included due to the fact they contain agents that were never approved for brain oncology, and likely their phase is not equal for all analyzed treatment arms. Approved drug targets are labeled as approved. The number of clinical trials is calculated by the number of trials, where a particular target has been drugged, each trial is counted only once, no matter the number of agents against the target. The number of unique drugs illustrates the number of drugs that exploit the target. For each drug, both pivotal and secondary targets were taken into account. The trend was calculated as follows: if a 3-year average of the target presence during 2017,2018, and 2019 is 10% or more higher than the average during the ten-year period – the trend is upward, whereas if it is 10% or more below than the average during the ten-year period – the trend is downward. The year column illustrates the first year the target appeared during these 10 years in brain cancer trials (excluding advanced biologicals).
- Supplemental Table 3 -
Supplemental Table 3 Drug target statistics (advanced biologicals). This table illustrates the whole list of targets for vaccines and advanced biologicals, like CAR T cells. The presence is calculated by the number of trials, where a particular target has been drugged. If in a single trial contains more than one agent against a particular target, this trial is counted times the number of these agents. Max phase column illustrates the maximal phase of a clinical trial, where a particular target has appeared. Trials with phase IV have not been included due to the fact they contain agents that were never approved for brain oncology, and likely their phase is not equal for all analyzed treatment arms. Approved drug targets are labeled as approved. The number of clinical trials is calculated by the number of trials, where a particular target has been drugged, each trial is counted only once, no matter the number of agents against the target. The number of unique drugs illustrates the number of drugs that exploit the target. For each drug, both pivotal and secondary targets were taken into account. The trend was calculated as follows: if a 3-year average of the target presence during 2017,2018, and 2019 is 10% or more higher than the average during the ten-year period – the trend is upward, whereas if it is 10% or more below than the average during the ten-year period – the trend is downward. The year column illustrates the first year the target appeared during these 10 years in brain cancer trials (advanced
biologicals).
- Supplemental Figure 1 -
