Both cancer cells and cancer stem cells exhibit metabolic reprogramming to survive and proliferate under unfavorable conditions such as hypoxia and chronic inflammation. Cancer stem cells show higher adaptation potential as compared with non-CSCs. You have not keep in mind, however, that it has been implicated that metastasis does not necessarily require EMT. I recommend you to read the following reviews in details.

Article Therapeutic Strategies Targeting Cancer Stem Cells

Article Metabolic reprogramming: The emerging concept and associated...

The heterogeneous expression pattern of CD44 and c-Myc is recognized at the invaive area of solid tumors characterized by breast and gastric cancer cells. The redox stress-induced canonical Wnt signal activation underlies this phenomenon.  

Article Inversed relationship between CD44 variant and c-Myc due to ...

I am including part of an unpublished paper that may help you answer your question:

Cancer stem cells metabolism differs from cancer “bulk” cells metabolism?

There is no clear answer to this question.

Ciavardelli et al (246) found that breast cancer stem cells (CSC) rely on Warburg effect (aerobic glycolisis).

Four seminal works by the team headed by Lisanti and Sotgia paved the way for a different concept.

In the first of these works (247) they used breast cancer cell lines in suspension and in attached monolayer. The mammospheres formed in suspension represent mainly CSCs, while those growing in monolayer represent “bulk” cancer cells.

(Mammospheres  origin comes from a tiny group of cells that exhibits stem cell phenotype and is able to grow in suspension culture and may originate tumors in mice.  Mammosphere grows in a three-dimensional shape and it does not grow in monolayer, while “bulk” cancer cell grows in monolayers).

They compared the proteomics of cells grown in mammospheres with those grown in monolayers, and they identified proteins that were selectively over-expressed in mammospheres. They found three groups of proteins that were highly over-expressed in mammosphere cells:

1)    mitochondrial enzymes

2)    proteins related with mitochondrial biogenesis

3)    proteins related with inhibition of autophagy or mitophagy

Based on these findings, the authors concluded that CSCs require accumulation of mitochondrial mass.

To test this idea they treated the cancer cells with inhibitors of monocarboxilate transporter 1 (MCT-1) and monocarboxilate transporter 2 (MCT-2) inhibitors. The inhibition of the transporters down-regulated the entrance of two mitochondrial fuels in cells: lactate and ketone bodies. The cells treated with MCT-1 and MCT-2 inhibitors showed a significant decrease in mammosphere formation which actually meant less CSCs.

These findings confirms previous publications by  the team headed by Lisanti and Sotgia regarding the “parasitic” metabolism in cancer (134), the second work we are considering here, where tumor cells have the capacity to obtain nutrients from normal host cells, such as fibroblasts. So that there is a group of cells in tumor stroma that predominantly uses aerobic glycolisis (Warburg effect) as source of energy and produces lactate that exits these cells. This lactate is incorporated by cancer cells generating energy through mitochondrial oxidative phosphorylation.  CSCs belong apparently to this last group and a decreased transportation of lactates through inhibition of MCT-1 and MCT-2 decrease the viability of CSCs.  

The third work to be considered (248) analysed 4 different groups of FDA approved antibiotics that target mitochondrial protein synthesis against 12 cancer cell lines (breast, prostate, melanoma, ovary, pancreas, brain, lung and DCIS). The antibiotics tested were azithromycin,  doxycycline (DOXY), chloramfenicol, and tigecyclin.

DOXY and tigecyclin showed dose dependent reduction in mammosphere production. DOXY inhibited tumor sphere formation with an IC50 between 2 and 10μM in breast cancer cell lines (MCF7 and T47D). At 200μM concentration of DOXY no mamospheres were formed. (2 to 5μM are clinically achievable concentrations, 200 μM is not ). When DOXY was tested with other tumor cell lines, it was also effective in all the cases, but the concentration used was 50μM which is higher than clinically achievable.

After a 100-mg IV dose, tigecycline serum concentrations are ∼1.5 μg/ml (Cuhna BA). With a high dose like 400-mg IV, a concentration of 6 μg/ml is attainable (Cuhna BA). The concentrations tested with this antibiotic in the experimental setting start with 10μM (approximately 5 μg/ml) which shows reduction in sphere formation but for an important reduction it is necessary to achieve a concentration of 50 μM/ml which cannot be obtained in the clinical setting.

In the fourth publication (249) they found that mitochondrial biogenesis is necessary for CSCs survival and propagation.

Tetracyclines  by interfering with oxidative phosphorylation and mitochondrial biogenesis may be useful tools against CSCs.

Metformin probably works in the same direction (134) and it may represent a synergistic pharmaceutical with tetracyclines.  There is strong evidence of metformin´s activity against CSCs (250- 258) and the mechanism of action is probably similar in this sense to that of tetracyclines: down-regulation of oxidative phosphorylation. This possible synergy deserves further experimental research.

Yang B et al (259) described the anti-CSCs activity of DOXY. Stem cell markers decreased with the treatment of these cells with DOXY and also invasion, migration, proliferation and colony formation were decreased. The importance of this research lays in a new method  to obtain these HeLa stem cells. HeLa-CSCs were treated with 20 μg/ml concentration of DOXY (this concentration is above clinically achievable concentrations). EMT (epithelial-mesenchymal transition) and all the stem cell markers were reduced in the treated cells. The expression of Snail and Twist were also significantly reduced.

The conclusion from the Lisanti and Sotgia group research, is that CSCs

1)   require an important mitochondrial mass,

2)   that the prevailing metabolic pathway is oxidative phosphorilation,

3)   that they can use lactate and other ketone bodies as energy supply (parasitic enslaving of associated fibroblasts) and

4)   down-regulation of phosphorylative oxidation (metformin or tetracyclines) or decreased fuelling of lactates (MCT1 and MCT2 inhibition) decreases proliferation and survival of CSCs. 

Possible explanations for these discrepancies:

1)   There are two different phenotypic cancer stem cells.

2)   There is a switch from an oxidative phosphorylation phenotype to a fermentative one at some point of cancer progression in the same way as in the “bulk” cancer cell.

3)   CSCs´phenotype is tissue or cancer specific.

4)   There are environmental factors, not identified yet, that may influence one or another phenotype.