Professional Documents
Culture Documents
Developing Treatment Guidelines For Myasthenia Gravis
Developing Treatment Guidelines For Myasthenia Gravis
ISSN 0077-8923
A N N A L S O F T H E N E W Y O R K A C A D E M Y O F SC I E N C E S
Issue: Myasthenia Gravis and Related Disorders
REVIEW
Address for correspondence: Donald B. Sanders, Department of Neurology, Duke University, DUMC Box 3403, Durham, NC
27710. Donald.Sanders@duke.edu
A task force of the Myasthenia Gravis Foundation of America recently published a formal consensus statement
intended to be a treatment guide for clinicians caring for myasthenia gravis (MG) patients worldwide. Its develop-
ment was stimulated by the fact that there is generally no accepted standard of care for MG, and no one treatment
is best for all MG patients. Also, there are few randomized trials of treatments in current use, and the generaliz-
ability of the few trials that have been successful may be difficult. Fifteen international experts in MG participated
in the consensus process, which used a simple consensus to develop preliminary definitions and the RAND/UCLA
Appropriateness Method to quantify agreement on treatment guidance statements for seven topics: symptomatic and
immunosuppressive treatment, intravenous immunoglobulin and plasma exchange, impending and manifest myas-
thenic crisis, thymectomy, juvenile MG, MG with muscle-specific tyrosine kinase antibodies, and MG in pregnancy.
The executive summary of the guidance statement was published with open access to facilitate access by patients
and healthcare professionals, and the full statement, with extensive background information, is available online. The
guidance statement is a living document that will require updates as new treatments and new information on current
treatments become available.
Keywords: myasthenia gravis; treatment guidelines; consensus process; RAND-UCLA Appropriateness Method
Introduction: why do we need myasthenia (MG). However, there is significant variation in the
gravis guidance treatment statements? treatment of MG, and there is no accepted standard
of care. This is mainly because large randomized
Several immunosuppressive and immunomodula-
controlled trials (RCTs) of MG therapies are few,
tory treatments are available for myasthenia gravis
and the generalizability of the available RCTs may be
a limited, largely because of the heterogeneous nature
These authors contributed equally to this manuscript.
†
Task Force panelists: Michael Benatar, University of
of the disease. Even the best RCTs cannot balance
Miami, Miami, FL; Amelia Evoli, Catholic University, all of the information needed to select the most
Rome, Italy; Nils Gilhus, University of Bergen, Bergen, appropriate treatment for a given patient. Hence,
Norway; Isabel Illa, Hospital de la Santa Creu I Sant an effort to develop consensus among international
Pau, Barcelona, Spain; Nancy Kuntz, Lurie Children’s experts was undertaken to guide clinicians on the
Hospital of Chicago, Chicago, IL; Janice Massey, Duke multifaceted approach to managing MG.
University Medical Center, Durham, NC; Arthur Melms,
University of Erlangen, Germany; Hiroyuki Murai, Inter- Background
national University of Health and Welfare, Narita, Japan;
Michael Nicolle, University Hospital, London, Ontario, Reviews of literature evidence are often used to
Canada; Jackie Palace, University of Oxford, Oxford, Eng- inform clinical practice. Narrative reviews are
land; David Richman, University of California, Davis, CA; descriptive reports that summarize the most impor-
Jan Verschuuren, Leiden University Medical Center, Lei- tant studies in the field, with perhaps some com-
den, the Netherlands. mentary about their clinical implications. Articles
doi: 10.1111/nyas.13537
Ann. N.Y. Acad. Sci. 1412 (2018) 95–101
C 2018 New York Academy of Sciences. 95
Developing treatment guidelines for MG Sanders et al.
included in a narrative review are typically deter- When good evidence is lacking, there is even more
mined by the authors and may be influenced by uncertainty about the value of differing options, and
their subjective opinions of the quality of the study, there is usually considerable practice variation.
availability of full-length articles, and other consid- When high-level evidence is not available to
erations. Therefore, these reviews may be subject to inform a CPG, guidance for practice can be based
selection bias. Moreover, when results of different on consensus methods, using the available literature
studies are inconsistent, with discrepant results, it and experience supplemented by the knowledge and
is often left to the reader to decide how to address opinions of expert clinicians. Group decisions are
these inconsistencies. based on the assumption that a group is less likely
Treatment guidelines are used not only by clin- than an individual to arrive at a wrong conclusion.
icians but also by third-party payers (insurance Groups can provide a broad range of knowledge
companies, Medicare, and Medicaid), governmen- and experience. Group interaction encourages
tal health organizations, and institutional review debate and scrutiny of multiple opinions, chal-
boards. In 2011, the National Academy of Medicine lenges preconceived opinions, and stimulates
(NAM), then known as the Institute of Medicine, new ideas. Group discussion therefore serves to
defined clinical practice guidelines (CPGs), often integrate judgments when uncertainty and differing
synonymous with evidence-based guidelines, as opinions exist, to identify areas of agreement, and
“statements that provide recommendations to opti- to determine areas where agreement is lacking.2
mize patient care, informed by a systematic review Informal consensus refers to the collective opin-
of the evidence and an assessment of the benefits ions developed in unstructured meetings or com-
and harms of alternative care options.”1 The NAM mittees with few formal rules or methods. Members
further established standards for development of in these groups are not instructed in any specific
CPGs, one of which is that they are developed from method, although simple directions may be pro-
a systematic review (SR) of all the available evidence vided, such as to not derogate input from other
to inform a clinical question. An important step in members.2
the SR process is measurement of the risk of bias of The following disadvantages of group consensus
each study that informs the CPG. Bias or systematic must be recognized: (1) only one person can speak
error is the tendency of a study to measure the effect at a time, limiting the number of ideas discussed; (2)
of the intervention on the outcome in a way that there is social pressure to agree with the majority or
deviates from the true value. This cannot be mea- a powerful voice; and (3) a desire to reach agreement
sured directly. However, principles of study design may result in premature closure without attention
can be used to estimate the risk of bias of a study. On to all options. Formal consensus methods have been
the basis of these principles, RCTs have the lowest developed to minimize some of these disadvantages
risk of bias, and case studies have the highest. by providing a structured decision-making process
There are several situations in which RCTs are not (e.g., by adopting rating methods to represent the
available, and it is not possible to develop an SR or extent of agreement). Performance in groups is
CPG because of the high risk of bias in the available affected in various ways by the presence of other
literature. High-quality studies are difficult to per- members. In formal consensus methods, there is a
form in some conditions and are therefore unlikely degree of anonymity in the process, which reduces
to be available. This is particularly true with rare the effect of wanting to please a senior or dominant
diseases like MG, where recruitment of sufficient member or to conform to judgments of the group.
numbers of subjects for clinical trials is difficult. An The group opinion is integrated, discussed, and used
additional factor in MG is that immunosuppressive to refine recommendations. A structured consen-
treatments require a long time to demonstrate effi- sus process, with formal rules and procedures, also
cacy. Financial constraints and study limitations, conveys authority and rationality to the ultimate
such as dropouts over time, may preclude high- product, leading to scientific credibility. The RAND-
quality studies, and the use of placebo controls may UCLA Appropriateness Method (RAM), developed
not be feasible or ethical. However, clinical and pol- in the 1980s by the RAND Corporation and the Uni-
icy decisions must still be made, and patients may versity of California Los Angeles, is one of the better
have questions about the options available to them. known methods of formal consensus.3
The following list of seven treatment topics to be input and discussion, we modified the statement
addressed was developed by the panel at the meet- and revoted.
ing: (1) symptomatic and immunosuppressive treat- The round 2 statement read: “A non-steroid
ment of MG; (2) intravenous immunoglobulin and immunosuppressive agent (IS) should be used alone
plasma exchange; (3) management of impending when steroids are contraindicated or refused. A non-
and manifest myasthenic crisis; (4) thymectomy in steroid IS should be used initially in conjunction
MG; (5) juvenile MG; (6) MG with MuSK antibod- with steroids when the risk of steroid side effects
ies; and (7) MG in pregnancy. is high based on medical co-morbidities. A non-
Draft guidance statements for each topic, along steroid IS should be added to steroids when steroid
with relevant literature reviews, were prepared by side effects deemed significant by the patient or the
the co-chairs and facilitator and sent to each panel treating physician, develop, response to an adequate
member for anonymous rating via email. Panel trial of steroids is inadequate, or symptoms relapse
members were asked to not discuss the topics upon steroid taper.” Round 2 voting resulted in a
directly with other members in order to reduce the median of 9 and a range of 8–9, indicating agree-
risk of undue influence by one or more panel mem- ment and appropriateness.
bers. The nonvoting facilitator tallied the votes and Here, round 2 was performed to address panel
summarized panel comments and discussion after comments and discussion and not because there
each round. Statements modified by the co-chairs was lack of consensus in round 1.
and facilitator on the basis of panel feedback were
then sent to the panel, along with the discussion for Thymectomy in childhood myasthenia
the next round of voting. This was repeated until In contrast, consensus on the topic of thymectomy
consensus was achieved; all statements achieved in childhood myasthenia required extensive dis-
consensus within three rounds of voting. cussion and modifications. The round 1 statement
Consensus on all topics was completed in May was: “In children and adolescents aged 5–10 years,
2015, and the final report was completed in July thymectomy should be considered only after fail-
2015. The executive summary of the guidance state- ure of symptomatic therapy and immunotherapy.”
ment was published in Neurology in August 20164 Round 1 voting produced a median of 6 and a range
with open access to facilitate access by patients and of 1–9, indicating that the statement was considered
healthcare professionals. A comprehensive version uncertain and that agreement was equivocal.
with extensive background information was pub- A modified round 2 statement was created on the
lished online. basis of discussion: “A. In patients under 15 years
of age, thymectomy should be considered in gen-
Examples of the consensus process eralized MG after unsatisfactory response to acetyl-
There was considerable difference in the ease with cholinesterase (AChE) inhibitors and immunother-
which consensus was reached among the different apy. B. There is wide consensus that thymectomy is
topics (Table 1). For example, consensus on the indicated in peripubertal and postpubertal children
topic of immunotherapy in MG was relatively with moderate to severe AChR-ab+ MG. C. Pub-
straightforward. lished reports also suggest that early thymectomy
(within the first 12 months of onset of symptoms)
Immunotherapy in MG is more effective than delayed thymectomy. D. For
The round 1 statement argued: “If high steroid doses seronegative children, there is always a risk that some
are needed chronically to achieve or maintain an will have a congenital myasthenic syndrome and
adequate response, a steroid-sparing agent should not immune-mediated juvenile MG. E. Evaluation
be added, typically along with the steroid, to per- at a center specializing in childhood neuromuscular
mit subsequent reduction of the steroid dose to the diseases should be considered before recommend-
lowest necessary to maintain an adequate response.” ing thymectomy in young patients with seronegative
Round 1 voting resulted in a median of 9 and a range MG.” Voting in round 2 resulted in a median of 8
of 6–9, indicating agreement and appropriateness. and a range of 2–9; four panelists rated it 2–4, and
Consensus on this topic was achieved after one the rest in the 7–9 range. Thus, there was still dis-
round of votes. However, on the basis of panel agreement and no consensus.
Table 1. Selected examples of consensus statements and voting rounds for each of the seven topics
Round 3
Statement Round 1 (median, range) Round 2 (median, range) (median, range)
I. Symptomatic and immunosuppressive
treatment of MG
1. Pyridostigmine should be part of the initial Multiple suggestions and Revote on combined, modified
treatment in most MG patients. consensus statements in this statement:
Pyridostigmine dose should be adjusted as round were combined into one 8, 4–9: appropriate, agreement
needed on the basis of symptoms. The ability statement in the next round.
to discontinue pyridostigmine can be an Individual statement votes:
indicator that the patient has met treatment 9, 2–9: appropriate,a agreement;b
goals and may guide the tapering of other 8, 5–8: appropriate, agreement;
therapies. Corticosteroids and/or IS therapy 8, 4–9: equivocal;
should be used in all MG patients who have 8, 5–9: appropriate, agreement;
not met treatment goals after an adequate 8, 2–9: equivocal;
trial of pyridostigmine. 7, 3–9: equivocal
2. Nonsteroidal IS agents that can be used in Multiple suggestions and Revote on combined, modified
MG include azathioprine, cyclosporine, consensus statements in this statement:
mycophenolate mofetil, methotrexate, and round were combined into one 8, 6–9: appropriate, agreement
tacrolimus. statement in the next round.
Individual statement votes:
8, 2–9: appropriate, agreement;
7, 3–8: equivocal;
7, 2–8: equivocal
II. Intravenous immunoglobulin (IVIg) and
plasma exchange (PLEX)
1. PLEX and IVIg are appropriately used as Multiple suggestions in this round Revote on combined, modified
short-term treatments in MG patients with were combined in the next statement:
life-threatening signs such as respiratory round. 9, 7–9: appropriate, agreement
insufficiency or dysphagia, in preparation for 9, 6–9: appropriate, agreement.
surgery in patients with significant bulbar
dysfunction, when a rapid response to
treatment is needed, when other treatments
are insufficiently effective, and before
beginning corticosteroids if deemed
necessary to prevent or minimize
exacerbations.
III. Impending and manifest myasthenic crisis
1. Although clinical trials suggest that IVIg and Multiple suggestions and Revote on combined, modified
PLEX are equally effective in the treatment of consensus statements in this statement:
impending or manifest myasthenic crisis, round were combined into one 9, 2–9: appropriate, agreement
expert consensus suggests that PLEX is more statement in the next round.
effective and works more quickly. The choice Individual statement votes:
between the two therapies depends on patient 8, 2–9: equivocal;
comorbidity and other factors, including 9, 6–9: appropriate, agreement
availability. A greater risk of hemodynamic
and venous access complications with PLEX
should also be considered in the decision.
IV. Thymectomy in MG
1. In nonthymomatous MG, thymectomy is Multiple suggestions and Revote on combined, modified
performed as an option to potentially avoid consensus statements in this statement:
or minimize the dose or duration of round were combined into one 8, 2–9: appropriate, agreement
immunotherapy, or if patients fail to respond statement in the next–round.
to an initial trial of immunotherapy or have Individual statement votes:
intolerable side effects from that therapy. 8, 1–9, equivocal;
8, 5–9, equivocal
2. The value of thymectomy in the treatment of Multiple suggestions and Revote on combined, modified Revote on modified
prepubertal MG patients is unclear, but consensus statements in this statement: statement:
thymectomy should be considered in round were combined into one 8, 2–9: equivocal 8, 7–9: appropriate,
children with generalized AChR-antibody statement in the next round. agreement
positive MG either (a) if the response to Individual statement votes:
pyridostigmine and IS therapy is 6, 1–9: equivocal;
unsatisfactory or (b) in order to avoid 6, 1–9: equivocal
potential complications of IS therapy.
Continued
Table 1. Continued
Round 3
Statement Round 1 (median, range) Round 2 (median, range) (median, range)
V. Juvenile MG (JMG)a
1. Maintenance PLEX or IVIg are alternatives to Multiple suggestions in this round Revote on modified statement:
IS drugs in JMG. were combined in the next 8, 6–9: appropriate, agreement
round.
8, 5–9: equivocal
VI. MG with MuSK antibodies
1. Many MuSK–MG patients respond poorly to 9, 6–9: appropriate, agreement
ChEIs, and conventional pyridostigmine
doses frequently induce side effects.
VII. MG in pregnancy
1. Provided that their myasthenia is under good Although consensus with Revote on minor modifications:
control before pregnancy, the majority of agreement was reached during 9, 7–9: appropriate, agreement
women can be reassured that they will this round, minor modifications
remain stable throughout pregnancy. If were suggested by the panel.
worsening occurs, it may be more likely 9, 7–9: appropriate, agreement
during the first few months after delivery.
2. Current information indicates that Multiple suggestions and Revote on combined, modified
azathioprine and cyclosporine are relatively consensus statements in this statement:
safe in expectant mothers who are not round were combined in the 8, 1–9: appropriate, agreement
satisfactorily controlled with or cannot next round.
tolerate corticosteroids. Current evidence Individual statement votes:
indicates that mycophenolate mofetil and 5, 2–9: equivocal;
methotrexate increase the risk of 8, 3–9: appropriate, agreement;
teratogenicity and are contraindicated 9, 6–9: appropriate, agreement
during pregnancy.
a A statement is considered to be appropriate if the median rating is 7–9 without disagreement, and inappropriate if the median rating
region containing the median score (i.e., most votes are in the same 3-point region that contains the median score).
A round 3 statement was generated by the dis- arrive at different conclusions. This is in contrast to
cussion: “A. The value of thymectomy in the treat- the evidence-based method for developing CPGs,
ment of pre-pubertal MG patients is unclear, but where the conclusions and recommendations will
thymectomy should be considered in children with be relatively similar across different panels, because
generalized AChR-ab+ MG either if the response to they are linked to the evidence from the SR. How-
AChE inhibitor and immunosuppressive is unsatis- ever, in the evidence-based method, a recommen-
factory, or if there is a need/desire to avoid poten- dation may not be possible if there is no high-level
tial complications of immunosuppressive therapy. evidence. It is therefore important for the reader to
B. For children diagnosed as seronegative general- be familiar with the method used to develop any
ized MG, the possibility of a congenital myasthenic practice guidance statement and to be aware of the
syndrome or other neuromuscular condition should strength and weaknesses of each method.
be entertained, and evaluation at a center special-
izing in neuromuscular diseases is of value before Looking to the future
thymectomy.” Round 3 voting produced a median
of 8, and a range of 7–9; indicating that a consensus This first international consensus guidance for MG
agreement statement was appropriate. management is not a static document; it will require
updates as treatments for MG continue to evolve.
In general, treatment recommendation statements
Limitations of consensus-based processes
should be reviewed every 2–5 years to determine
The major disadvantage of consensus-based state- whether an update is needed.1,5,6 The NAM does
ments is that the recommendation depends on the not provide a timeframe for updates but recom-
makeup of the expert panel and the personalities mends regular review of the literature to determine
involved. Two groups with different experts may whether new evidence dictates a needed change in