Quality Evaluation of the Clinical Trials for Natural Products Used in Cancer: An Evidence-Based Literature Review

The amount of data regarding the use of herbs/herbal products in cancer clinical trials at times creates a great challenge for oncologists to prescribe or counsel patients. It urges critical evaluation of the quality of clinical trials. Herein, for the first time, the clinical trials for herbs used in cancer were critically evaluated on the basis of three widely used scales, i.e., Jadad, Delphi, and Cochrane scales. The literature was collected with the help of online databases, journals, libraries, and books using a number of specific keywords as mentioned in detail in forthcoming sections. A total of 73 clinical trials were extracted, evaluated, and scored for 14 herbs, according to the predefined criteria mentioned below. A major deficiency of “non-blinding of clinical trials” was observed. The principal component analysis revealed four components (PC1–PC4) with a total variability of 68.21%, wherein the highest percentage variability was observed for PC1 loaded with “non-blinding of the clinical trials, no concealment of the treatment allocation, non-blindness of the patient and care provider”, which accounted for 30.81% of the total variability. The next major variability of 14.70% was observed for PC2 loaded with “non-randomization of the studies, non-blinding of the outcome assessors, no proper drop-out procedures, and lack of information regarding baseline characteristics for the groups”. Pearson’s correlation further confirmed a similar correlation pattern for the mentioned deficiencies (p = 0.05). An in-house grading scale was developed, showing a very small portion (16.44%), i.e., 12/73 studies with a good quality, whereas the majority (57.54%) of the studies, i.e., 42/73, were found to be of poor quality. The rules and regulations governing the quality of clinical trials needs to be more stringent and updated for the natural products/herbs used in cancer clinical trials.


Introduction
Cancer is a term implicated for an uncontrolled cell division that may invade nearby tissues and spread to other body parts via blood and lymph systems [1]. The major risk factors for this disease include age, family history, hormones, tobacco use, irradiations, chronic inflammation, diet, and sedentary lifestyle [2,3]. A study reported 1,735,350 cancer cases in the United Stated for the year 2018, with total deaths of 609,640 [4]. The new estimates for 2040 revealed a global burden of up to scale. Furthermore, an aggregate score was calculated and the quality of clinical trials was agreed on the basis of an in-house developed rating scale.

Inclusion Criteria
Inclusion criteria included clinical trials published in the English language; clinical trials using natural products and studies/reported in human subjects; any clinical trials using natural products with established folklore uses in cancer or with a reported use in a community worldwide (ethnopharmacological relevance); any clinical trial (phase I-V) reported for natural products in cancer, irrespective of blinding, randomization, statistical models, outcomes, and results presented; and all clinical trials using natural products along with conventional medication.
For ethnopharmacological relevance, a list of herbs/natural products was sorted and evaluated individually for its reported use in cancer in any community worldwide. The information regarding ethnopharmacological or folklore uses was searched in reputable journals and any data presented in the form of interviews, community surveys, or data collected from local inhabitants/herbal practitioners was extracted and analyzed.

Exclusion Criteria
Exclusion criteria included clinical trials reported for cancer using sources other than natural products; any clinical trial for cancer using natural products without proper ethnopharmacological relevance or community use in cancer; all natural products with a sound ethnopharmacological relevance in cancer but are yet to be evaluated in a clinical study; all preclinical studies (in vivo animal models or cell culture reports); any incomplete or duplicate study; clinical trials using vitamins, minerals, and conventional drugs only; and phase-0 clinical trials.
Review period: An extensive search strategy was applied where retrospective data were collected without any restriction from September 2019 to April 2020. The literature was collected according to eligibility criteria, studied, and reported in the review herein. Until preparation and finalization of the manuscript, we updated the literature data on a regular basis, and any new information, if obtained, was added to the literature search.
Search result: The literature search consisted of 1342 articles, which was confined to 73 following a proper scrutiny of the eligible articles according to the pre-defined criteria. The flow diagram for selection and scrutiny of the literature is given in Figure 1.

Literature Search
The relevant literature data were studied and the extracted information is presented in a step-wise pattern below.

Folklore Uses/Ethnopharmacological Relevance of the Selected Herbs in Cancer
The ethnopharmacological relevance for the included herbs is given in detail in Table 1. The herbs were reported to have folklore uses in cancer in various communities including those of Pakistan, Palestine, Morocco, India, Turkey, Bangladesh, Ghana, Jordan, and Yemen. Various parts of these herbs such as leaves, fruit, dried sap, bulb, flowers, barks, rhizomes, seeds, rind, resin, and oleogum have been used for the intended purposes. Various types of cancers such as breast, head, skin, stomach, colorectal, liver, lungs, esophageal, and prostate cancers have been cured or treated using these herbs. These applications were used as a source of evidence for folklore or ethnopharmacological relevance of the selected herbs in order to evaluate the application and quality of clinical trials performed.
The current section highlights the key features of each clinical trial conducted for herbs in cancer. The part of an herbal product used as such or in a dosage form, the type of cancer studied, and the observed outcomes in a clinical trial were concluded as shown in Table 2.

Evaluation of Clinical Trials (Jadad, Delphi, and Cochrane Scale)
Three different scales, i.e., Jadad, Delphi, and Cochrane, were used in this study to evaluate clinical trials individually as per the items mentioned in the scales (Table 3). Each scale evaluated the quality of a clinical trial on certain specific key features mentioned in the table, where most of the items were overlapping/common for the three scales. Every scale possessed positive as well as negative aspects in terms of evaluation criteria and it is worth mentioning that no "best-fit scale" exists to critically analyze a clinical trial; hence, the three scales were applied together for assessing the quality of each clinical trial. This step may help cover the deficiencies present in a scale. In addition, as Jadad is the only scale that uses a scale (0-5) to assign score to a study, whereas Delphi and Cochrane lack a scoring system. For the sake of simplicity and ease of calculation/comparative scoring, we assigned an internal score to each item of the Delphi (0-9) and Cochrane scale (0-10). This summed up the total score for the three scales as being between 0 and 24 [18]. All the clinical trials included in this study were assessed and scored using the three scales and an individual as well as final score were assigned for classification of the clinical trials. Table 4 shows in detail all the items present in the three scales and the deficiencies observed for individual clinical trials according to the items in each scale.  powder capsules breast cancer no effect in chemotherapy-induced nausea/vomiting [77]     Was the compliance acceptable in all groups?
8. Did the analysis include an intention-to-treat analysis?
Was the drop-out rate described and acceptable? Was the timing of the outcome assessment in all groups similar?
Wormwood N1 X X X X -1 X X X X X X X -X −1 X X X X X X X X --−1 −1 * Jaded deficiencies: a: randomization mentioned, b: randomization method, c: double-blind words, d: double-blind method, e: description of withdrawals and dropouts. ** Delphi scale deficiencies: a: randomization performed, b: treatment allocation concealed, c: similarity at baseline, d: eligibility criteria specified, e: outcome assessor blinded, f: care provider blinded, g: patient blinded, h: point estimates and of variability presented for the primary outcome measured, i: intention-to-treat analysis. *** Cochrane scale deficiencies: a: randomization adequate, b: treatment allocation concealed, c: similarity at baseline, d: patient blinded, e: care provider blinded, f: outcome assessor blinded, g: co-interventions avoided or similar, h: compliance acceptable, i: description of withdrawals and dropouts, j: similarity in timing of the outcome assessment.

Statistical Analysis
Statistical tools of PCA (principal component analysis) and Pearson's correlation were used to categorize the data. The factors analyzed showed a total variability of 68.21% for four components (PC1-PC4), as shown in Table 5. The scree plot for the components is presented in Figure 2. An individual variability (%) was observed as PC1 (30.81), PC2 (14.70), PC3 (12.69), and PC4 (9.99). The factors loaded in PC1, i.e., the highest variability, were the deficiencies of non-blinding of the clinical trials, no concealment of the treatment allocation, and non-blindness of the patient and care provider. The next high percentage variability, i.e., PC2, showed loading for deficiencies, non-randomization of the studies, non-blinding of the outcome assessors, no proper drop-out procedures, and lack of information regarding baseline characteristics for the groups. The individual percentage variability with cumulative percentage is shown in Table 5. In addition, a three-dimensional representation of the deficiencies in the components is shown in Figure 2.

Statistical Analysis
Statistical tools of PCA (principal component analysis) and Pearson's correlation were used to categorize the data. The factors analyzed showed a total variability of 68.21% for four components (PC1-PC4), as shown in Table 5. The scree plot for the components is presented in Figure 2. An individual variability (%) was observed as PC1 (30.81), PC2 (14.70), PC3 (12.69), and PC4 (9.99). The factors loaded in PC1, i.e., the highest variability, were the deficiencies of non-blinding of the clinical trials, no concealment of the treatment allocation, and non-blindness of the patient and care provider. The next high percentage variability, i.e., PC2, showed loading for deficiencies, non-randomization of the studies, non-blinding of the outcome assessors, no proper drop-out procedures, and lack of information regarding baseline characteristics for the groups. The individual percentage variability with cumulative percentage is shown in Table 5. In addition, a three-dimensional representation of the deficiencies in the components is shown in Figure 2.  Pearson's correlation was constructed to cross-verify the PCA analysis. For Pearson's correlation, all the pairs showed a positive correlation, i.e., none of the pairs were found to have a negative correlation. A similar phenomenon to PCA was observed in Pearson's correlation. Even for the deficiency "co-interventions were not mentioned" that was loaded separately in PC4 of the PCA was observed with no correlation/pair to any other factors in Pearson's analysis. The correlation matrix for Pearson's analysis is shown in detail in Table 6.
The statistical analysis below confirms a descending order (Figure 3) of occurrence for the deficiencies.  Pearson's correlation was constructed to cross-verify the PCA analysis. For Pearson's correlation, all the pairs showed a positive correlation, i.e., none of the pairs were found to have a negative correlation. A similar phenomenon to PCA was observed in Pearson's correlation. Even for the deficiency "co-interventions were not mentioned" that was loaded separately in PC4 of the PCA was observed with no correlation/pair to any other factors in Pearson's analysis. The correlation matrix for Pearson's analysis is shown in detail in Table 6.
The statistical analysis below confirms a descending order (Figure 3) of occurrence for the deficiencies.   "non-blinding of clinical trials > no compliance reported for the study > co-interventions not mentioned > outcome assessor non blinded > intention-to-treat analysis was not included in a study > treatment allocation was not concealed > patients were non-blinded > care provider was non-blinded > group baseline characteristics were not mentioned > studies were non-randomized > drop-out procedure were not mentioned > timing for outcome assessment were not mentioned".

Score for Clinical Trials
An in-house grading scale was developed in order to simplify the classification of quality of the bulk of clinical trials. The clinical trials were graded out of 24 points, with distribution as "very poor quality" (≤6 including negative values), "poor quality" (7 to 12), "acceptable quality" (13 to 18), "good quality" (19 to 24). Detailed information regarding study score is given in Table 7. A small portion, i.e., 16.44% (12 out of 73) of the studies were found to be of good quality, whereas more than half the proportion (57.54%; 27 very poor + 15 poor quality) of the studies were found to be of poor quality.

Discussion
The evaluation and assessment of the clinical trials developed a few basic questions necessary for any study such as "what is the source/background of the herb?", "what is its phytochemical profile?", "what are the parts, dosage forms, and extraction solvents used?", "what is the asking information, dose used, and its clinical phase studied?" All these questions are summarized briefly below.
Source and background data regarding the herb: Basic information relevant to the herb in terms of family, genus, species, folkloric use in cancer, geographical origin, and identification from authentic resources are very important. The plant may vary with regard to phytochemistry, which is subjected to differences in terms of place of origin, wherein altitude, temperature, stress, salinity, irrigation, etc. may affect the nature and quantity of active chemical present [110,111]. The majority of the trials were unable to explain the authentication process for the source of plant used and its background information. This may affect the quality of a clinical trial.
Phytochemical profile for the part used: The part of a plant may differ in the nature and amount of active chemicals when compared to other parts of the same plant; hence, there is a need for proper phytochemical profiling. These clinical trials used various parts of the plants such as leaves, roots, fruits, infusion, juice, and essential oils; however, the phytochemistry for the part of the plant used was missing in most of the studies [100].
Final dosage form used and its preparation/extraction: A number of clinical trials used dosage forms (extract or dried powder in capsule/tablet, creams, gels, mouthwashes, liquid extract, pills, syrup, etc.), however, the method of extraction or dosage form preparation was observed often. For a herbal product to be effective, appropriate extraction/processing is the basic step for success. The factors involved in extraction/drying/preparation of final dosage form (sunlight, temperature, solvent polarity and non-polarity, extraction time, pressure etc.) may either degrade or enhance the amount of an active ingredient or its activity thereof [112][113][114]. It is very crucial to investigate the effect of these factors upon the potential of an herbal dosage, but none of the clinical trial undertook such an investigation.
Choice of green solvent and extraction: Most of the extraction at present is performed using green solvents (water, ethanol, acetonitrile, etc.) due to dual properties of being eco-and human-friendly with less adverse effects. The infusion, juice, and fermentation products prepared in the reported clinical trials used alcohol-based solvents that are unhealthy, costly, and carry more adverse effects. Most of these solvents release toxic chemicals upon heating [115]. A need to shift to green extraction may be promoted.
Phase (0-V) selection: Although phase-0 is mandatory for sub-therapeutic dose and toxicity, often the phase-I studies are skipped for the herbs with proper ethnopharmacological/folklore data available at community level [116]. In such cases, phase-II studies are started without a prior phase-I study, and hence proper evidence is necessary for an herb to start with a clinical trial.
Masking of clinical study: Equally important, a clinical trial should be properly blinded with respect to the patient, care provider/administrator, and data assessor in order to avoid the risk of bias in the data [117,118]. A number of clinical trials we reported do present the issue of improper blinding.
Treatment strategy (interactions, complications, and duration): The duration of treatment needs to be shorter in order to avoid complications, particularly in subjects using conventional medication for treatment. These clinical trials continued the studies from weeks until years, which is difficult at times because the subjects enrolled may either have stopped conventional medication or are already using natural products. Herbs are best known for their cytochrome P450 inhibitory or induction properties, of which both are dangerous. A long-term treatment strategy may expose the subjects to various herb-drug/drug-food interactions and nutritional deficiencies, which may produce emergency conditions. All these factors are the major sources of non-compliance in a study, being was observed in most of these clinical trials [111,119,120].
Dose used: A dose up to 10 g in most of the cases was observed in these clinical trials. It is quite difficult to administer such a high dose in the form of a tablet or capsule as it outweighs the capacity for available size. In addition, it becomes impossible to administer a huge dose in divided doses, especially in subjects using conventional drugs where a serious risk of herb-drug interaction exists. More importantly for herbs with a lack of phase-0 data, it is a serious risk to use such a high dose that can predispose potential health risks. This urges researchers to explore the herb for proper phase-0 data, half-life, PKs (Pharmacokinetics), PDs (Pharmacodynamics), etc. and to ensure the quality variation for active principle in herbs, prior any clinical study [121,122]. The reported clinical trials did not mention any such information, which is utmost required for a study.

Recommendations to Enhance Quality of a Clinical Trial
Ethno-botanical/pharmacological and folklore evidence with quality evaluation: A detailed literature search needs to be ensured in order to collect appropriate information regarding the folklore use of a herb in various communities, followed by uniformity of geographical origin, part, family, genus, and species of the herb to study. In addition, a proper phytochemical profile must be established to evaluate and declare the quality and quantity of active chemicals present in a herb that are responsible for anticancer effects [117,118].
Herbal pharmacovigilance: Herbal pharmacovigilance is necessary to ensure the mainstream data necessary for a herbal clinical trial. Phase-0, i.e., toxicity studies, sub-therapeutic dose selection, adverse effects of the herb/herbal product, long-term effects, and herb-drug/food interaction studied, as well as PKs and PDs for half-life, metabolizing enzymes, and the excretion process need to be established. Pharmacovigilance assures a small dose with shorter treatment strategy/duration so as to avoid the untoward effects of the herb, especially when combined with conventional drugs. In addition, the pharmacovigilance ensures the quality variation and standardization of herbs [121,122].
Need for extraction or isolation and shape of final dosage form: The researcher needs to be clear regarding the pros and cons related to extracts/extraction and isolation. It is tiresome to isolate an active chemical; however, extracts due to presence of multi component nature pose restriction and complications when it is needed to study the molecular or genetic level effects for a treatment. Pursuit of isolation of the main active ingredient responsible for the cancerous effects in a herbal product/extract is mainly favored. In addition, dosage form compatibility is more important. Powder drugs and injectables are more easily absorbed and show enhanced bioavailability and therapeutic effects; still, the idea of nanoformulations, i.e., nano-particles, emulsions, micelles, and gels are more preferred due to low dose, enhanced and targeted treatment, and less side effects or adverse effects.
Clinical trials with tailored treatment approach: The paradigm shift from empirical to tailored approach, i.e., treatment strategy based on biologically relevant question, is the upcoming future for cancer clinical trials. Although the idea demands a profound change in infrastructure and methodology of clinical research and is challenging, it will bring about new opportunities in cancer treatment with a better understanding of the disease and mechanism of action of the treating agent [123]. The clinical trials may focus on acquiring such concept.
Immune-oncology: Clinical trials with a specific focus on immune boosting properties are also a unique source of accelerating cancer treatment [124]. A number of herbs such as leaves of muicle, aguacate, and muerdago; bark of cuachalalate and una de gato; and roots of matarique and guizazo de caballo have immune-enhancement/stimulant properties that may serve as a novel source of cancer treatment [125].
Nano-oncology: Nano-dosage form in the shape of liposomes, dendrimers, gold nanoparticles, micelles, nanoemulsions, nanogels, etc. are widely used in cancer treatment as they are inert, noncorrosive, targeted, safe, and free of the adverse effects associated with conventional treatments. A number of nano-dosage forms such as Myocet and Doxil for doxorubicin are available in the market [126]. It is worthwhile to convert the herbal products or extracts into various nano-dosage forms, which may add the benefits of more therapeutic efficiency and less adverse effects.
Precision medicine: Avoiding the idea of "one-size-fits-all" and matching the most appropriate and relevant individualized treatment approaches for a patient on the basis of the genetic profile of the patient and cancer type is known as precision medicine [127] In spite of tumor heterogeneity, which may affect precision medicine, promising outcomes may be observed if precision medicine is applied in herbal clinical trials.

Conclusions
The clinical trials in the systemic review revealed a poor quality according to the evaluation scales used. The majority of the studies were non-blinded and non-randomized. With respect to herbs used, a proper pharmacovigilance background was not reported in the studies. It is highly recommended that researchers enhance/uplift the studies of these clinical trials via addition of appropriate ethnopharmacological relevance, quality variation and standardization, phytochemical profile, focus on the hot area of cancer, and precision medicine when planning to conduct a clinical trial using a herb (powder/extract, etc.) or herbal product.
Author Contributions: R.A. and L.H.A. conceived the idea, study design, methodology, statistical analysis, discussion, and conclusion; L.H.A., A.K.A. and S.M.A. was responsible for the literature review, data extraction, introduction write up, evaluation, and scoring of individual clinical trial with tables for ethnopharmacological relevance and deficiencies/scoring of the clinical trials, as well as arrangement of the references. R.A. conducted analyses. All authors have read and agreed to the published version of the manuscript.

Funding:
The study declares no funding from any governmental/private organizations.