3 Our hypothesis
The next two sections provide background
material about human starvation and ketogenic dieting that is necessary for understanding our hypothesized treatment which
we detail in Section 3.3
3.1 Starvation: Minimizing the Body's
The body’s glucose requirements are minimized during periods of starvation, which constitutes
drastic transformation in the body’s energy metabolism. Broadly speaking
starvation has five stages which are outlined in the Table 1 below.
Normally by the end of the first day, hepatic glycogen stores are depleted and the liver
(and to a lesser extent, the kidneys) begin to produce glucose from pyruvate, glycerol and amino acids. The vast majority
of tissues in the body (such as skeletal muscle, the heart, and most organs) are facultative in their choice of substrate
for energy production. During this period of glucose scarcity, they cut their glucose metabolism and increase their metabolism
of free fatty acids (FFAs). Starvation even at this early stage forces body tissues
to switch from glycolysis to lipid oxidation for their energy needs. The ketone bodies Beta OHB, acetoacetate and acetone
are always present in the blood, but after a few days of starvation they begin to be produced in large quantities in the liver.
Most tissues of the body can metabolize ketones (with the exception of acetone).
Stages of starvation
Decreased gluconeogenesis and increased cerebral ketone consumption
Table 1. The five stages
of starvation from the time of last ingestion (reprinted from Cahill, 1983, page 2).
Of special interest is stage 5--prolonged starvation of two weeks and longer. Cells have long since
cut their uptake of glucose but now also cut their uptake of ketones. This change,
coupled with increased blood-brain-barrier permeability of ketones (Morris, 2005) we will call keto-adaptation." During
stage 5, most cells rely purely on FFAs for their metabolic requirements (Robinson and Williamson, 1980). However, some tissues still retain a need for glucose even after keto-adaptation. The largest consumer is the brain and CNS, whose health is vital to the success of our hypothesis.
How much glucose does the brain need? A typical brain
glucose requirement (Cahill, 1970, Figure 1). This is a preferential utilization
but not an obligatory utilization. This point is crucial to our hypothesis. During keto-adaptation, the brain glucose requirement drops
to approximately 44 g/day (Cahill, 1970, Figure 5).
What supplies the energy that this difference of approximately
100 g/day of glucose previously supplied?
[The brain consumes about 150 g/day of glucose.] The ketone bodies -OHB and acetoacetate
in a 6:1 proportion (Owen et al., 1967, Table 5). Note that the brain does not metabolize
FFAs (Owen et al., 1967, Table 4). In stage 5 starvation, ketone blood levels
surge to 2.5-9.7 mmol/L which are in contrast to a negligible level of .01 mmol/L in non-starving subjects. The brain will readily use ketone bodies if they are available even in non-fasting periods, but under this
condition ketone usage is negligible (Sokolo , 1973). Morris (2005) suspects that the brain's ketone
metabolism was evolutionarily developed to ensure survival in times of glucose being scare.
Cahill (1970) suspects that other body tissues slash their ketone consumption
in favor of FFA's to ensure adequate substrate for the brain. Further, Cahill and Veech (2003).
The fuel metabolism of stage 5 starvation is illustrated in Figure 1. Of special interest is the inventory of the sources of gluconeogenesis that provide the 44 g/day CNS requirement: stored triacylglycerols
(15g), amino acids taken from muscle tissue and to a lesser extent other tissues (20g), and pyruvate / lactate (14g),
which is the end product of anaerobic glycolysis (thus about a third of the glucose during starvation burned by the CNS is
recycled back into glucose via the Cori cycle).
The figure of 24 hours Basal Metabolism (~`,500
CAL) won’t copy
Figure 1: Fuel sources and sinks
in the daily metabolism of Stage 5 starvation assuming a
1500 calorie resting metabolism (inspired by Cahill, 1970, Figure 5).
Starvation can be a useful tool to minimize glucose needs in all but a few body tissues. Fasting, while relatively safe (Stewart and Fleming, 1973), is not only unpleasant but leads to muscle
wasting and weight loss, both of which are serious concerns in cancer patients. How
do we minimize the body's glucose needs while avoiding these undesired effects? We turn to a solution in the next section.
3.2 Keto-adaptation via the 4:1 Ketogenic
A means to emulate the metabolic conditions of starvation while still ingesting food is the same
engineering problem faced in the 1920's by the expert childhood epileptologist Dr. Russel M. Wilder. He observed that short-term starvation was therapeutic for children with
intractable epilepsy, but obviously impractical as a long-term solution. He innovated the “ketogenic diet" (KD) that mimics the characteristic starvation metabolism we outlined
in Section 3.1. It is still used today to reduce seizures (for an historical account
and a recent clinical trial, see Neal et al., 2008).
[RM] Wilder's KD is a high fat, low protein and very low carbohydrate diet (McDonald, 1998). In what is called the “4:1
KD," carbohydrate must be limited to 20 g/day
(the amount the muscle tissue provides during prolonged starvation) and protein must be limited to an amount to offset normal
(non-starvation) protein turnover, estimated at about 25
g/day for a 70kg person (National Research Council (US) Food & Nutrition Board, 1989). Any additional protein has the potential to be gluconeogenic and thereby should be avoided. All other calories must be supplied by fats.5 Commercial
formulas for this diet are readily available. [High fat, butter on cellar/lettice with 25 gm of protein diet for fast—Jk
Are the physiological markers in a prolonged 4:1 KD exactly the same as markers under prolonged
starvation? Depending on the macronutrient proportions, serum glucose can be slightly higher and ketone levels slightly lower
(Phinney et al., 1983). We assume that a 4:1 KD is equivalent to keto-adapted metabolism observed during prolonged starvation
described in Section 3.1: glucose uptake minimization, increased cerebral ketone uptake, and exclusive use of FFAs for energy
production in every other tissue in the body. Thus, a KD can achieve starvation metabolism adaptation without the unpleasant
side effects of hunger and muscle loss.
Simone et al. (2013) also reviews
other benefits of the KD such as muscle sparing and possible even muscle synthesis which can be beneficial for cancer patients
with risk of cachexia. The KD is generally well-tolerated with minimal
side effects such as constipation, salt loss, mild acidosis and increased incidence of kidney stones, when employed chronically"
and has been demonstrated to be safe even in patients with advanced cancer (Schmidt et al., 2011). Kossoff et al (2007) reports on the longest duration of a KD: 21 years with virtually no side effects. There is also some
evidence that ketones in their own right are directly toxic to cancer (Magee et al., 1979; Skinner et al., 2009; Fine et al.,
2009) and can improve the immune system's ability to target cancer (Husain et al., 2013). This can be an added
benefit of the KD that can work in tandem with our hypoglycemia strategy.
Our hypothesized treatment
begins with a cancer patient being administered a 4:1 KD for more than 14 days to allow for keto-adaptation. Then our proposed
acute treatment is begun. We turn to implementation details now.
3.3 Our Proposed Treatment
Section 2 discussed previous literature that experimented with both starvation and the KD for cancer
therapy via glucose starvation. We believe these studies did not reduce serum
glucose to levels sufficient to starve cancer. On a KD, “plasma glucose concentration fall only mildly, remaining in the normal range in normal-weight individuals
... [thus] ... simple tumor glucose starvation is therefore unlikely in humans" (Fine et
Serum glucose in prolonged starvation drops to about 65mg/dL (Cahill, 1970, Table 2) and the prolonged
4:1 KD can also result in a similar serum level, but it is unclear why further reduction in the blood glucose level should
not be similarly well-tolerated after keto-adaptation.
In an adult, cerebral blood flow is approximately 750mL/min = 0.13 dL/s which implies 8.1mg/s of
glucose can potentially be delivered to the brain. In Section 3.1, we noted that the brain has an obligatory requirement of
44 g/day = 0.51 mg/s6 which is only 6% of the glucose available at a serum concentration of 65 mg/dL. Although experiments
evaluating the glucose extraction ratio in human subjects are limited, animal studies suggest that the brain is able to extract
40% of the serum glucose in the hypoglycemic state (McCall et al., 1986). Thus, we theorize that the body is being conservative.
Serum glucose concentrations below the level where 0.51 mg/s can be actively transported into neurons for mere moments can
result in CNS damage; 65 mg/dL is comfortably above this threshold.
We have evidence for this conjecture that keto-adapted
subjects can survive on serum glucose much lower than 65mg/dL. Hypoglycemia has been reported in therapeutic fasting or starvation
for over 100 years. Chakrabarty (1948) investigated 407 starvation cases in Bengal famine to find that 20 or so of them had
chronic blood sugar below 40 mg/dL with no symptoms of hypoglycemia… We have evidence for this
conjecture that keto-adapted subjects can survive on serum glucose much lower than 65mg/dL. Hypoglycemia has been reported
in therapeutic fasting or starvation for over 100 years. Chakrabarty (1948) investing rated 407 starvation cases in the Bengal
famine to and that 20 or so of them had chronic blood sugar below 40 mg/dL with no symptoms of hypoglycemia. Stewart and Fleming
(1973) who supervised a world-record-setting 382-day therapeutic fast found that blood glucose became stable at 30 mg/dL,
dropping intermittently to 20 mg/dL without symptoms. The most relevant evidence to our hypothesis is the observation of intentional
acute hypoglycemia induction by Drenick et al. (1972) who keto-adapted obese patients via 45-60 days of starvation and then
perfused insulin. Within the hour, they observed serum glucose as low as 9 mg/dL without acute hypoglycemic symptoms. Serum
glucose this low would normally induce coma (which has even been observed with blood sugar as high as 40-49 mg/dL according
to Ben-Ami et al., 1999, Figure 3).
Our plan is simple. After keto-adaptation with the 4:1 KD, we propose to acutely attenuate the serum glucose
concentration to severely hypoglycemic levels such as observed in Drenick et al. (1972) on a patient-by-patient basis. Since
insulin is required by cancer cells and fuels their metabolism (Iqbal et al., 2013) and also can impede ketogenesis, we choose
not to induce hypoglycemia via insulin perfusion (as in Drenick et al., 1972), but instead propose the use of a gluconeogenesis
One possible inhibitor is the diabetic drug metformin which inhibits gluconeogensis indirectly. Metformin may
have other anti-cancer effects (Jalving et al., 2010, Figure 3) and there is certainly
supporting epidemiological evidence for metformin's therapeutic effect in cancer (Ben Sahra et al., 2010). For these reasons,
it has been proposed for use in a hypothesized treatment similar to ours (see Oleksyszyn, 2011 and further in Oleksyszyn et
al., 2014). However, metformin has many drawbacks. First, it is not a potent inducer of hypoglycemia in the doses that are
normally administered. Metformin is also slow acting and not rapidly titratable. Additionally, the drug activates adenosine
monophosphate activated protein kinase which may have other effects on metabolism under keto-adaptation (see Jalving et al.,
2010, Figure 2). Lastly, supratherapeutic doses have been associated with fatal lactic acidosis.
Due to these shortcomings, we propose using a drug which is a pure liver-kidney gluco-neogenesis inhibitor. Luckily,
these drugs have been investigated by the diabetes community where high endogenous glucose production is considered one of
the characteristics of the disease and is therefore a therapeutic target.
There are many control points within gluconeogenesis. The unidirectional synthesis of fructose-6-phosphate from
fructose-1,6-bisphospate is a prudent point to interrupt because it is removed from glycogenolysis and mitochondrial function
(thus, it would have the fewest side effects). The synthesis is controlled via the rate-controlling enzyme fructose-1,6-bisphosphatase.
Two inhibitors that have been clinically evaluated are named CS-917 and MB07803 (van Poelje et al., 2011). They have good
safety profiles and can flexibly attenuate
gluconeogenesis to any desired level.
Returning to our discussion of the target level of hypoglycemia, we comment that we do not know the effective
dose (ED) curves for serum glucose levels for tumor necrosis or the lethal dose (LD) curves for human body death after keto-adaptation.
It is possible that a 9 mg/dL glucose concentration may be too high to be toxic for the cancer. If this proves to be the case,
further hypoglycemia would need to be safely induced for treatment efficacy. We turn to this now.
Since the brain is the only tissue with an obligatory glucose requirement, we opt for direct perfusion of glucose
into the brain intra-arterially via the carotid arteries at the rate necessary to provide minimum brain glycolytic needs.
Perfusion of glucose directly to the brain has been performed successfully in rats (Borg et al., 1997) and baboons (Conway
et al., 1969). Once the brain receives its required glucose, we propose an increase in the dose of gluconeogenic inhibitor
drug to attenuate serum glucose to a negligible level which is safe enough for erythocytes, blood marrow and leukocytes to
survive. The brain would perceive normoglycemia but everywhere outside of the brain the serum would be extremely
hypoglycemic. If necessary, we can also perfuse _-OHB in tandem if there
is risk that the level of ketosis is not sufficiently high for CNS metabolic needs. This protocol would require extensive
proof-of-concept work in animal models before any attempt to undertake study in human subjects.
The reader may wonder if this strategy would be effective against brain cancer since the brain is still receiving
glucose. Seyfried et al. (2003) reports data supporting that such a treatment may be effective since glucose uptake in brain
cancers drops with lower serum glucose and that the cancer cannot metabolize ketones, which may even be toxic to them. How fast could this procedure induce tumor death? We echo Oleksyszyn et al. (2014),
“we do not have any idea as to how deep hypoglycemia needs to be to trigger spontaneous remission." All estimates are
drawn from data from in-vitro studies and thus are not generalizable to human prognoses.
Spitz et al. (2000, Figure 2) reports between 55% and near
100% necrosis rates in three cancer lines within 48hr after total glucose deprivation. Complete withdrawal is not
possible in the human body for reasons explained above and our treatment could deliver minimum serum levels estimated at 5
[They propose direct infusion of glucose via the carotid artery and block glucose in the body with
a drug similar to metformin. I doubt the need for this radical intervention.] This protocol would require extensive proof-of-concept
work in animal models before any attempt to undertake study in human subjects. … Spitz et al. (2000, Figure 2) [in vitro]
reports between 55% and near 100% necrosis rates in three cancer lines within 48hr after total glucose deprivation.
We propose a therapy for starving cancer by inducing keto-adaptation followed by the induction of therapeutic
hypoglycemia via gluconeogenesis inhibitor drugs. We then propose rescue cerebral glucose infusion through direct catheterization
of the brains's blood supply. This would be followed by an increase of the gluconeogenic
inhibitor drug to lower serum glucose even further in non-cerebral tissues.
A main concern is that each person is a unique metabolic entity" (Seyfried et al., 2014). The calculations presented throughout this manuscript are meant to be illustrative
of an average person. For each individual, personalized dosing and administration schedules will require iterative fine-tuning.
Proper monitoring and its associated costs are a concern. Hypoglycemia is very risky but might not necessarily
require the monitoring setting of an intensive care unit (patients with diabetic ketoacidosis are frequently treated with
insulin infusion outside of the intensive care unit). Although the costs of the diet and the medications needed to induce
hypoglycemia in our proposal might be far lower than the cost of novel chemotherapeutic agents, the cost of close monitoring
may substantially erode this benefit.
There are several other concerns with speculative treatment. Cancer is heterogeneous and few predictors are currently available as to an
individual tumor’s response to hypoglycemia. There is research suggesting
that some cancer cells can make use of ketones and FFAs. Not all cancer lines
proliferate rapidly and intense anabolism is not mandatory (Jose et al., 2011. Our hypothesized treatment would not work in such cancer lines. It is not known whether the proposed hypoglycemic
levels can be sustained for long periods of time without symptoms. Starvation research suggests that hypoglycemia is safe
in the long term, but these accounts documented serum levels many times above our proposed serum level and they were predominantly
documented in obese subjects without cancer. An important consideration is whether keto-adaptation additionally protects against
the neuro-hormal cardiovascular and electrophysiologic consequences of hypoglycemia (Frier et al., 2011). Additionally, glucose
is not the sole nutritive substrate for cancer cells. For instance, glutamine can also serve as a
major metabolite (Yuneva, 2008), as well as fermentation from other amino acids. There are times when the body is in need of glucose such as during inflammation and
infection which limits the setting of our hypothesized treatment.
There is also a fair chance that our proposed treatment is more effective when provided in tandem
with other therapies. For instance, Abdelwahab et al. (2012) demonstrated that brain cancer in mice was curtailed and did
not recur with the simultaneous administration of the 4:1
KD and radiation therapy. Allen et al. (2013) demonstrated that the 4:1 KD enhances radio-chemo-therapy in mice
models of lung cancer. See Klement and Champ (2014) for a discussion of how the KD can act synergistically with radiation
therapy. Also, Lee et al. (2012) demonstrated that fasting enhances chemotherapy. Poff et
al. (2013) demonstrated that hyperbaric oxygen therapy is enhanced by the KD.
Alternative adjuvant therapies may also be fruitful. For example, Majumdar et al. (2009) demonstrated
that the combination of curcumin with resveratrol could be an effective therapy in colorectal cancer, Singh and Lai (2004)
demonstrated cancer apoptosis from wormwood extract
and Tin et al. (2007) demonstrated that astralagus saponins
can inhibit colorectal cancer cell proliferation. There are many other such supplements that can be administeredsimultaneously
with our hypoglycemia treatment.
We would like to thank Justin Bleich and Stephen Kapelner for helpful discussions and comments on this manuscript.
We would like to thank Marie Le Pichon for illustrating Figure 1. Adam Kapelner
acknowledges support from the National Science Foundation's Graduate Research
Fellowship as well as support from the Simons Foundation Autism Research Initiative.
Taken from references and examined:
Eliot & Barnett 2011 not up 9/20/15) because of website modifications
on 1000 patients must obtain from UCSD http://onlinelibrary.wiley.com/doi/10.1111/j.1528-1167.2008.01853.x/pdf
Schmidt on KD diet terminal patient proved safety, but was too short to measure change in tumor size. Allow 70 gm carb/day. http://www.biomedcentral.com/content/pdf/1743-7075-8-54.pdf
Seyfried 2008 http://onlinelibrary.wiley.com/doi/10.1111/j.1528-1167.2008.01853.x/pdf