Mitochondrial diseases are a group of heterogeneous disorders caused by dysfunction of adenosine triphosphate (ATP) synthesis and insufficient energy sources. ATP is mainly produced by the mitochondrial respiratory chain and is regarded as a molecular energy source. Thus, mitochondrial dysfunction is more significant in high metabolic organs/tissues, such as the central nervous system (CNS), heart, muscle, liver, and kidney. The CNS is one of the principal organs susceptible to mitochondrial diseases, as it is the latest organ in tissue differentiation, lacks self-repair ability, has a high energy demand and low ATP storage capacity, and is prone to mitochondrial function disorders. The incidence rate of epilepsy in mitochondrial disease ranges from 35 to 60%, and one-third of patients with refractory epilepsy have biochemical evidence of mitochondrial dysfunctions (1).

All types of mitochondrial diseases have impaired ATP production; therefore, patients present with clinical symptoms of energy failure, such as weakness, hypotonia, reduced physical stamina, muscle atrophy, heart failure, and limitation of eye movements with ptosis. Short stature, reduced muscle bulk, deceleration of head growth, and other developmental issues are observed with mitochondrial diseases as well (2). The etiology of mitochondrial disease is primarily related to genetic factors involving mutations in the nuclear DNA (nDNA) and mitochondrial DNA (mtDNA) (3). The mechanisms of epilepsy in mitochondrial diseases may be related to decreased ATP levels, abnormal calcium absorption, energy exhaustion, and energy resource shortage (4–6). Some investigators believe that mitochondrial diseases and epilepsy are closely correlated, creating a vicious cycle (6).

Diagnosis of mitochondrial disease can be confirmed in some patients by the identification of mtDNA mutation with a test of blood sample assessment, and clinical symptoms can be used to infer some disorders. However, many patients require more systemic diagnostic methods, such as family history, blood and/or cerebrospinal fluid lactate concentration, neuroimaging, cardiac evaluation, and gene testing for mtDNA mutations and nDNA mutations (7). Applications of “cocktail treatment” and anti-seizure medications (ASMs) are still mainstream and first-line approaches to treating mitochondrial diseases with epilepsy. Seizures in mitochondria, including generalized tonic-clonic seizures and status epilepticus, can be treated with traditional ASMs, but few of them work well. Moreover, the utilization of traditional ASMs may affect the function of mitochondrial respiratory chain enzymes or have direct toxicity for mitochondria. For instance, valproic acid may impair mitochondria and liver function, cause teratogenesis, and inhibit carnitine absorption. Some studies have reported that the use of valproic acid may lead to severe results like sudden and fatal liver failure (8–11). In summary, it is a challenge to effectively control seizures and enhance energy metabolism in mitochondrial diseases. Therefore, we propose a ketogenic diet (KD) as a possible treatment for mitochondrial diseases with epilepsy.

The KD is a high fat, low carbohydrate, and balanced protein diet, which is used in the treatment of glucose transporter-1 deficiency syndrome, pyruvate dehydrogenase deficiency (PDHD) syndrome, epilepsy, tumors, and some inherited metabolic diseases. In 2015, the KD became a routine treatment for epilepsy in China (12).

The KD simulates the metabolic consequences of fasting, in which the body utilizes fat as an energy source instead of carbohydrates. Some studies have demonstrated that the KD is effective against mitochondrial diseases, such as epilepsy caused by mtDNA and nDNA defects. Recently, reports of gene studies on mitochondrial disease have begun to explain the mechanisms at work. Apart from the direct anticonvulsant function of medium chain triglycerides, other mechanisms include the dysregulation of fat metabolism, downregulation of carbohydrate metabolic genes, improvement of mitochondrial biogenesis, and regulation of anticonvulsant function by peroxisome proliferator-activated receptor-γ (PPAR-γ) and astrocyte metabolism (13). Additionally, PPAR-γ also seems to be involved in the anticonvulsant effect of anticonvulsant drugs (14–16). Therefore, the use of the KD to treat mitochondrial diseases with epilepsy is theoretically feasible.

Lee et al. investigated the effect of the KD in 24 patients with mitochondrial respiratory chain enzyme deficiency who also had refractory epilepsy. They found that 75% of participants had >50% seizure reduction, and half of the participants were seizure-free (17). A retrospective study of KD application in refractory epilepsy, caused by respiratory chain enzyme deficiency, showed that after KD intervention, seven participants (50%) were seizure free with three having no relapse, one having >90% seizure reduction (not seizure free), and three having 50%−90% seizure reduction (18). Some case studies have reported the application of the KD in other mitochondrial diseases: treated patients had satisfactory therapeutic effects and good tolerance for POLG mitochondrial disease, PDHD syndrome, or respiratory chain complex I deficiency due to Otahara syndrome (19–21).

Although some studies have illustrated the safety and efficacy of KD in mitochondrial disease treatment, most of these were case or small-sample studies, and there have been no relevant studies in China. We conducted a multicenter clinical observational study of KD treatment in patients with mitochondrial epilepsy to confirm the efficacy of KD in the treatment of mitochondrial epilepsy. We aim to summarize the conditions of mitochondrial epilepsy treatment with KD in China, further confirm the efficacy and safety of KD, and help improve the quality of life of the patients.

Our results showed the benefits of KD in mitochondrial disease treatment. Few studies have discussed the efficacy of KD in the treatment of mitochondrial diseases. A recent systematic review reported that 7/8 patients with mitochondrial diseases responded to KD treatment (5 seizure-free, 1 reduction of seizures, and 1 control of status epilepticus) (23). Two other studies with mitochondrial respiratory chain defects in epilepsy patients found that 71–75% of participants experienced >50% seizure reduction after KD intervention (17, 18). The response rates of mitochondrial diseases in this study were similar to those in refractory epilepsy studies, in which response rates for the KD and control groups were 35–56.1% and 6–18.2%, respectively (24–31). In addition, the International League against Epilepsy (ILAE) also attests to considerable response rates (70%) of KD to epilepsy and recommended this therapy in its guidelines (32). On comparing our results to previous studies and ILAE guidelines, with the 76% (19/25) response rate of trial-completed participants, it is reasonable to state that KD could effectively improve seizure conditions of mitochondrial diseases with epilepsy.

Surprisingly, the response rate of the KD group seemed to be lower (although this was not statistically significant) than that of the control group who had changed to KD after 1 month of the control period. However, the explanations for this could vary. First, eight participants in the KD group withdrew, while only two withdrew from the control group after changing to KD. A high withdrawal rate certainly influenced response rates. Second, although most participants in both groups started with a 2:1 ratio diet, more participants changed to a lower ratio in the KD group after dietary adjustment. Some studies have noted that the classic KD with a higher ratio may lead to better seizure control (33). However, other studies have indicated that there is no difference in efficacy between high and low ratios (34). The relationship between the ketogenic ratio and treatment efficacy remains controversial. Another plausible reason for the difference that we reported is participant bias. Even though statistical analysis found no difference in seizure frequency between groups at baseline, it seemed that participants in the KD group had relatively more seizures.

Other than reduction in seizure frequency, some participants experienced favorable biomarker changes and cognitive improvements at 3 months. Similar biomarker changes have also been demonstrated in studies of mitochondrial disease (35, 36). The Neuropsychological Development Scale (0–6 years old) and Wechsler Scale showed that three participants had a development quotient and mental age increase, or total score improvement. The results of cognition improvement in our study were not as good as those from epilepsy treatment (37, 38), probably because only a few participants underwent cognitive assessment.

MELAS is a maternally inherited multisystemic disorder caused by mutations in mitochondrial DNA and is the second largest subgroup of diseases in this study. The relationship between certain gene mutations and MELAS has been extensively discussed. One study showed a high consistency between MELAS and m.3243 A>G variants (39). According to that study, the m.3243 A>G variant in the mitochondrial tRNA gene was present in 21/23 MELAS participants, 11/11 oligosymptomatic relatives, and 12/14 asymptomatic relatives, but not in 5/50 mitochondrial disease (39). Another study indicated that gene variants were found in 26/31 participants with MELAS, but only in 1/29 participants with CPEO (chronic progressive external ophthalmoplegia), 0/5 participants with MERRF (myoclonus epilepsy with ragged-red fibers), and 0/50 control participants (40). In general, more than 80% of participants with MELAS have the m.3243 A>G variant, with m.3271T>C, the second most common mutation, found in 10% of participants (41, 42). In our study, all participants with MELAS were diagnosed with the m.3243 A>G mutation. In addition to the m.3243 A>G variant, some other gene mutations presented in this study were also found to be related to mitochondrial diseases in previous studies, such as m.3697G>A and m.8344A>G (43–45). To the best of our knowledge, mutation of m.127720 A>G and m.15596 G>A, have not previously been discussed and our study is the first to relate them to a particular disease.

In our study, KD was found to be highly effective (89%) in participants with MELAS. However, the relationship between MELAS and KD has not been studied extensively. In a cellular model, exposure of neuronal cells to ketone bodies increased ATP synthesis, improved mitochondrial metabolism, and restored the activity and stability of complex I, which is impaired in MELAS (46). In a case study, a 22-year-old female patient's seizures were unaffected by the addition of new anti-epileptic medications; however, after receiving a modified KD, she experienced an improvement in seizure control and a decrease in stroke-like episodes (47). From a genetic perspective, KD strongly effects mitochondrial DNA mutations, especially the m.3243 A>G variant, which was the largest gene mutation subgroup. An earlier study indicated that a severe m.3243 A > G mutation might impair fuel catabolism and cause MELAS (48). Some studies have proposed that KD may induce electron transport chain subunit mRNA to increase mitochondrial biogenesis. This results in increased ATP levels, leading to increased neuronal “energy reserves,” allowing neurons to withstand metabolic challenges and stabilize neuronal membrane potentials (49, 50). KD appeared to be effective for other mtDNA mutations in this study, including m.3697 G>A, m.8344 A>G, m.3243 A>G+m.3271 T>A, m.127720 A>G. The response rate of nDNA mutations was much lower, as only four participants with COQ4, POLG, and PDHB variants were participants. In nDNA-related research, low glycemic index treatment (LGIT) was found to be effective in treating mitochondrial epilepsy with POLG1 gene mutation. In the report, the visual aura and aphasia could not be resolved in an adult participant with three ASMs. After the introduction of the LGIT, the headaches, aphasia, and visual aura progressively improved and disappeared. The participant improved and returned home two weeks later, and the seizures disappeared (20).

Ten participants withdrew and 7/10 reported poor compliance. Most of these had trouble with food operations and feeding. The high-fat and low-carbohydrate food composition was quite different from the regular diet, especially in China, where carbohydrate rich foods such as rice are still the predominant component in a meal. Thus, education and training for participants or caregivers prior to dietary intervention is important. Some researchers believe that new therapies are urgently needed to broaden the management options and improve the prognosis of patients (51), thus, tools of e-health management and recipe software could have increased the treatment compliance (52, 53). Additionally, correct management of the KD in patients with refractory epilepsy is important from the beginning to avoid side effects (54). One participant died during the study period. Epilepsy was noticed in this participant 2.5 months after birth, and he was diagnosed with COQ10D7. His eating ability was poor; therefore, constant tube feeding was used. At 11-months of age, he was admitted to the KD group and passed away 1 month later. He had normal stools and marked seizure improvement, but experienced vomiting. No other medical condition was present. The cause of death was unknown and may have been related to sleep apnea. According to previous reports, respiratory distress, cerebellar atrophy, and lactic acidosis are common symptoms of COQ10D7.

In total, seven participants experienced side effects during the intervention, with five reporting gastrointestinal (GI) side effects, including vomiting, bloating and GI disturbance. This is similar to other studies (26, 55–57) showing that approximately 30%−40% of participants had GI side effects, but our study showed a lower incidence. This difference could account for the relatively low ketogenic ratio observed in our study. One participant reported hyperlipidemia, which is not unusual in KD, but some studies have found that long-term KD intervention often keeps the serum lipid profile within normal limits (58, 59).It is important to manage side effects properly to encourage patients to remain on the KD.

To the best of our knowledge, this is the first prospective controlled study to discuss the efficacy of KD in the treatment of mitochondrial diseases with epilepsy. The larger sample size in this study made the therapeutic effects of KD on mitochondrial diseases more credible. Detailed genetic information for each patient also confirmed that KD was more efficient in MELAS and mtDNA mutations, especially for the m.3243 A>G pathogenic variant. However, our study has some limitations. The first limitation was the brief period of the normal diet in the control group, however a full 3-month period without dietary treatment may not be acceptable to parents. The second limitation is the unbalanced number of participants in the KD and control groups. Third, the use of parental seizure records might have introduced subjective errors such as overestimation or underestimation of seizure numbers. The combined utilization of subject records and EEGs may improve the scientific weight. Long-term blind trials are suggested for further studies.

Our study showed that KD is effective and can be included in the management of patients with mitochondrial disease and epilepsy. Gene reports indicate that KD is more efficient for MELAS and mtDNA mutations, especially the m.3243 A>G pathogenic variant. The side effects are considerable but should not be excessive. Most side effects were easily manageable, resulting in a few patients withdrawing from diet intervention.

The original contributions presented in the study are included in the article/Supplementary Materials, further inquiries can be directed to the corresponding authors.

The studies involving human participants were reviewed and approved by the Medical Ethics Committee of Chinese Clinical trial registry, with a Clinical trial registration No. of ChiCTR1900020789. Written informed consent to participate in this study was provided by the participants' legal guardian/next of kin.

DS, FF, JL, and LH designed the study and revised the manuscript accordingly. The other authors participated in the study and provided case information. MY, MF, and DS drafted the manuscript. All authors contributed to the article and approved the submitted version.

This study was supported by the Fund of Futang Research Center of Pediatric Development (No.FTCSF-2018-05), the Fund of Shenzhen Zeneca Biotechnology Co., Ltd (No.20181208), and China Association Against Epilepsy (CAAE) Research Fund—Qitong Fund (No.CJ-B-2021-20). Shenzhen Zeneca Biotechnology Co., LTD. was not involved in the study design, collection, analysis, interpretation of data, the writing of this article or the decision to submit it for publication.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.