With the 2018 Farm Bill, the United States legalized cannabis which contained low levels of THC, reversing in part the Controlled Substances Act of 1970, which had made most all Cannabis illegal even for medical use. 

Despite the 1970 Act which falsely stated that the plant had no medical use, the United States government, at the University of Mississippi, grew and provided tons of Cannabis to pharmaceutical companies from 1968-present.

It is my opinion that the government and pharmaceutical companies have not been faithful in sharing their taxpayer-funded research on Cannabis with the public.  This sentiment influences my decision to propose an open-source, volunteer model.  I see this Model as a simple, effective and legal way to work around conflicts of interest and regulations.  Although I feel it is necessary to disclose my sentiment, I will not discuss it further in this paper as I do not wish to detract from the Model.

The proposal has three main sections:  Medical Theory, the Volunteer Model, and Statistical Theory.  All of it is meant to be a general outline.  The point here is to show the public a way to do things yourself without having to rely on the government and pharmaceutical companies to do it for you.  

“It’s promising, but it needs more research.”  I read that in all types of papers on herbs.  This paper shows a way that anyone can test their theories.


Disclaimer:  I have no medical qualifications.  In addition to consulting with your physician, I recommend that people who are interested in this subject read the sources in this section thoroughly and do further research on your own.  In this section I will provide a brief summary of my review of literature and some opinions. 

Warning:  Cannabis root contains Epifriedelanol and Friedelin.  Another species that contains Friedelin, Mayentus ilicifolia (Espinheira Santa) has been used by “native groups in Paraguay, where women use the plant as a contraceptive and fertility regulator, and to induce menstruation and abortions.”  (Source). Ryz, Remillard and Russo mention “Friedelin may have estrogenic activity.”  All of this is to warm women who are pregnant or may become pregnant.  The medical properties of the roots are largely untested and unknown, except for historical use.

Historical Anti-Tumor Usage of Cannabis Roots

Russo Ryz, Remillard and Russo in Cannabis Roots: A Traditional Therapy with Future Potential for Treating Inflammation and Pain provided a history of the medical use of cannabis root.  They referenced several old medical journals in the 18th Century that said the roots were good for tumors (although the researchers could not tell if the “tumors” the old journals described would have been cancerous). 

Epifriedelanol and Friedelin in Cannabis Roots

Ryz, Remillard and Russo wrote:  “There is currently no research available about the activity of friedelin or epifriedelanol specifically isolated from cannabis roots.”  

The Journal of Pharmaceutical Sciences, however, published a study in December 1971, Chemical constituents of cannabis sativa L. Root, which specifically states:  “An extract of Cannabis sativa L. root yielded two pentacyclic triterpenes, friedelin and epifriedelanol, and N‐(p‐hydroxy‐β‐phenylethyl)‐p‐hydroxy‐trans‐cinnamamide. Structures of the triterpenes were confirmed by preparation of derivatives and comparison to authentic friedelin. Epifriedelanol was also oxidized to friedelin.”  Although It is unclear to me if the Cannabis roots in the study published by Journal of Pharmaceutical Sciences were grown at the University of Mississippi, the “Details-  Funding Information” section of the article lists “Research Institute of Pharmaceutical Sciences, School of Pharmacy, University of Mississippi.”

Screen Capture of

Seattle Weekly’s summary of the 1971 study:  “Researchers also discovered friedelin, an antioxidant heralded for its ability to promote liver health. The most intriguing discovery was a compound called epifriedelanol, which demonstrated tumor-killing effects.”

Other Compounds in Cannabis Roots

There isn’t much data on some of the compounds other than Epifriedelanol and Friedelin which are present in Cannabis roots. These other compounds identified by Ryz et al are listed below, with a link to their Wikipedia page, if one is available.

Carvone. Dihydrocarvone. Cannabisativine. Anhydrocannabisativine. Sitosterol. Campesterol. Stigmasterol. N-(p-hydroxy-β-phenylethyl)-p-hydroxy-trans-cinnamamide. Choline.

Why Breast Cancer?

Four reasons:  1. Historical Anti-Tumor Usage. 2. The relative ease of monitoring the cancer.  3.  Topical efficacy.   4. Estrogen regulating potential. 

Historical Anti-Tumor Usage. As previously referenced, Cannabis Roots have been used for treatment of unknown tumor types for centuries.

Monitoring:  Although the compounds in Cannabis Roots may have wide-ranging applications for cancer and other diseases, Breast Cancer is a relatively easy Cancer to monitor with mammograms, and that’s a major factor in a volunteer model that doesn’t have funding.

Topical Efficacy:  The idea to test Cannabis Root compounds on Breast Cancer was inspired, in part, by reading a study on D-Limonene and Breast Cancer, which used a topical application of D-Limonene.  “We hypothesized that limonene would distribute extensively to human breast tissues.”  The goal of a topical application is to achieve a good concentration of the active compounds in the affected areas without having to run the compounds through your entire body.  Cannabis roots, historically, seem to do this quite well. 

Estrogen:  I don’t want to be too specific in this section because I don’t have the qualifications to do so.  Suffice to say that Breast Cancer has a lot to do with estrogen receptors.  The oldest and highest-selling Breast Cancer drug in the world, Tamoxifen, is an estrogen receptor modulator. Tamoxifen was first developed for birth control in the early 1960s.

Ryz et al reference studies that have to do with the positive effects of Friedelin on estrogen-related functions in mice.  The plant Mayentus ilicifolia (Espinheira Santa), which contains Friedelin, has been used for estrogen-related purposes.  Ryz et al said “Friedelin from the stem bark of Mesua daphnifolia (a threatened species) had weak cytotoxic activity against four cancer cell lines, including MDA-MB-231 (human estrogen receptor-negative breast cancer).”


This model depends on volunteers to come together online or by mail in one data collection project.  

Producers and Product

I believe that roots will one day be grown indoors to bring out certain compounds and qualities, as it is grown today to bring out properties in the flowers.  Initially, though, I think it is best to begin studies with roots in their most natural state available.  In order to grow a good root mass that has a good tap root, and bring out any natural compounds in the root that are dependent upon solar radiation, I recommend that the plants should be grown outdoors and in the ground.  

The soil should be tested first to make sure it is free of any pollution.  Pesticides and Fertilizers should be avoided.  Cannabis plants are strong and hardy.  Who cares if the bugs eat a little bit.  These plants should be grown for their roots.

There are potentially many ways to create a product from the roots.  I think the simple way it has been done for centuries is a good starting point.  Fresh roots should be rinsed in water to remove all dirt.  The roots should then be pressed or crushed somehow.  Next, boil the roots.  Some compounds will float to the top.  Add dairy butter, which will combine with these compounds and float, and then skim off the butter.

Hopefully some producers will donate some product for those who want to try it.  It’s for a good cause.  If it works, you’ll be selling plenty of it later.

Why butter?  In a topical application, your looking for a substance that interacts well with tissue and penetrates the tissue without rash, discomfort, etc.  Combining the compounds in the roots with butter or oil is the way the substance was used over the centuries.  Butter seems to be a substance well absorbed by the skin, which is important so that the root compounds can reach the area of tumor underneath the skin.  While there are certainly other oils and so forth which may also work or even work better, butter just seems the best first choice.  If anyone wants to use a substance other than butter, it is no matter statistically, as MLR will sort the data. In fact, having multiple preparations will highlight what works best.  Further and again, all that I say in terms of medicine is merely a suggestion.  I value feedback and other’s opinions. The general concept of how to analyze the data is all I hope to make clear.

I see no reason to favor growth of one Cannabis variety over another at this early stage of research.  It will be statistically helpful to have many varieties of product.  The variance in compounds produced by multiple producers across multiples Cannabis subspecies, geography, soil type, etc., will make the model statistically stronger.  The variance between compounds in samples will highlight what compounds do what.  It is only necessary to record the exact sample composition used by the volunteer patient.


Patients will be on their own resources to provide data on both the change in tumor size and the other relevant data including change in tumor size.  I envision the patients receiving the products by mail, using the product, and recording their change in tumor size over time along with the other relevant data. They can upload their data to a website or mail it in and let someone else do it. Hopefully some medical facilities will volunteer here.

Administration of Model

Administration volunteers will manage the aggregation of data into a spreadsheet.  This data will then be uploaded to a public source for research by anyone who is interested.


Once available, college grad students looking for a good data set, or anyone else for that matter, can run whatever model they feel like.

The statistical model I envision is Multiple Linear Regression using Ordinary Least Squares.  I’m no statistical guru, but that’s the one I would try.  It’s simple in concept and application.  It works.  It appeals to the Central Limit Theorem. 

Clinical Testing versus Multiple Linear Regression

In a clinical study, the goal is to isolate one dependent variable against one independent variable.  Cancer is the dependent variable, dependent on the treatment. Sufficiently similar groups are divided into test and control groups.  The test groups have to be similar enough so that the differences between the individuals in the test group don’t affect the outcome.

The control group receives a placebo.  The test group(s) receive treatment. In theory, by eliminating all other things which could have caused a change in the dependent variable, only one thing remains to cause any change- the treatment. 

Multiple Linear Regression

With Multiple Linear Regression, you can test an infinite number of factors which could cause or cure cancer simultaneously.  And there are, of course, multiple factors in the real world which cause and cure cancer. You can probably think of many- your age, medical history, genetic factors, etc. In a regression we would say that your cancer is a function of these factors.

Conceptually, think of a spreadsheet. Rows and Columns. Each row is a different patient (observation). The first column of the spreadsheet is Change in Tumor Size. It is the dependent variable. Every other column is a factor that can be identified which influences the size of the tumor. These factors are things like your age, medical history, and of course the medication taken which is being tested. This big spreadsheet filled with plenty of observations is called a matrix in statistics. MLR uses matrix algebra to “invert the matrix.” This process tells you both the statistical likelihood that each independent variable influenced the tumor size, and the degree to which it did so.

You are only limited by what you can identify and quantify.   Groups need not be sorted to achieve sufficient homogeneity for the clinical testing model.  Those factors which would otherwise separate patients are instead collected as data.  None are excluded from the study.  Patients can take other medication, smoke, or whatever, as long as you record what is relevant. (Which is not to say that it is a good idea to mix medicines or to smoke, just that MLR can sort the data without problem). All receive the test medicine.  

The clinical testing model obviously can’t handle that many independent variables simultaneously. The more you throw at an MLR model, the stronger it gets. “Adding independent variables to a multiple linear regression model will always increase its statistical validity, because it will always explain a bit more variance.” (Source)

So if the MLR model is so superior statistically to the clinical testing model, why isn’t it used by the government (FDA, NIH, NCI, etc) and pharmaceutical companies? Good question. I won’t try to answer the question here. I will add just a bit of history on MLR and science.

Carl Friedrich Gauss

MLR and OLS trace their origins back to the Gauss-Markov Theorem and the work in the 20 or so years previous to the theorem being published in the early 19th century. Carl Friedrich Gauss gets most of the credit. Gauss developed his OLS/MLR method to explain the elliptical orbits of planets. He had identified six factors which explained their orbits. (Source). He called the method his “gift to the world” that could explain all sorts of things.

The use of MLR in medicine was likely first discussed by Hilda Mary Woods and William Thomas Russell of the London School of Hygiene and Tropical Medicine in 1931 in their book An Introduction to Medical Statistics. Austin Bradford Hill, of the same London school, followed up with Principles of Medical Statistics in 1937. Hill had received a degree in Economics in 1927. (Source)

Data to Collect

I propose to collect data on Demographics, Genetics, and Medical. Demographics are things like Age, Gender, Body Mass Index, etc. Genetics, I’m not sure how to describe, but perhaps genetic factors which influence cancer can be divided into classes 1, 2, 3, etc. Medical is very wide ranging obviously. Medical to me are things like smoking, medication and herbs you are taking, or other diseases you may have. Add a column for anything relevant. Take smoking for example. Did you smoke this week? If yes 1, if no 0. Or you could smoking it into 3 classes, none, light, or heavy. You get the picture. Same for any medication, disease, or any other relevant medical factor. Just quantify the factors and record them. Some researchers may omit data if they choose to, and some data sets may not be complete, but in general, it is better to have more data than less.

Privacy concerns should be addressed by stripping out all publicly identifiable information from the data set, although we all know in this era that privacy can’t be controlled completely.

It is very important that the compounds in the cannabis roots be identified in their exact concentrations. These compounds are all independent variables. Concentrations of the compounds will vary, and so the results will vary. If the product works, the variance in results will show exactly how it worked. Can one factor influence another? Of course. Let’s say that accidentally, we discover that a drug or herb you are taking interacts well with a compound in the roots and cures the cancer. Some people call a discovery of this nature “serendipity.” MLR can show you a lot of serendipity. In Econometrics, running a big model without a theory beforehand is called data mining, and it is frowned upon. I see no reason not to do so in this application however. While knowledge without understanding can be dangerous, this is an open source model, and all the scientists in the world can later explain any processes that involve two or more compounds.

Serendipity: Pileated Woodpecker in Flight was not the Original Subject


Forget the statistics for a second. I hope that Cannabis Roots will work so well on Breast Cancer that the results will be positive across the board and undeniable. If it works, the type and amount of data that I recommend will be plenty useful to understand how the product worked.

The only caveat to a volunteer model is safety. If you want to try an unknown product, you’ll need to do a lot of research first to understand the risk. It is your right to try. If the product doesn’t work, that doesn’t mean that the model doesn’t work. You can test any herb against any disease in the manner outlined here. My best shot to cure a disease with an herb is Breast Cancer with Cannabis Roots. What’s yours?


I have a friend, Shannon, who informed me he had been recently diagnosed with cancer. Shannon was convinced that Cannabis Roots held the key to cure cancer, based on what he said was “third world” use of the herb to treat cancer. Having some years ago lost my mother to cancer, using the status quo treatments of chemotherapy and radiation, I was left with a feeling that there had to be a better way. I offered to research the topic of Cannabis Roots and Cancer for Shannon and let him know the results. This paper is the result. I believe he is correct. Shannon also grows Hemp and hopes to develop a product from the roots. I do not have a financial interest in developing a product with him, but we have discussed the possibility of it in the future. Like Gauss, I’m giving this Model to the world. I don’t care who profits from this as long as Cannabis Roots get tested in an honest way that doesn’t involve the government and pharmaceutical companies.