Salmon Calcitonin Use and Associated Cancer Risk

Introduction

Calcitonin was discovered in 1961, with the first synthesized salmon version created in 1969.^1-3 Since creation, calcitonin has found a role in clinical medicine and has been approved by the US Food and Drug Administration (FDA) for use in the treatment of postmenopausal osteoporosis (PMO), hypercalcemia, and Paget’s disease.^4-6 Although calcitonin can be synthesized from humans and animals alike, salmon calcitonin (SCT) has primarily been used in pharmaceutical applications because it is 40 to 50 times more potent than human calcitonin, has analgesic qualities that are useful in the treatment of bone and other sources of musculoskeletal pain, and has a low toxicity profile.^3,6,7 Calcitonin is a polypeptide that participates in calcium and phosphorous metabolism and reduces bone turnover. In randomized clinical trials, SCT has shown the ability to reduce bone turnover and increase bone mineral density (BMD) when compared with placebo in postmenopausal women.^8-11 However, data from randomized controlled trials (RCTs) published since calcitonin’s approval have shown that other agents have superior efficacy for fracture reduction, whereas calcitonin’s ability to reduce fracture remains unsubstantiated.^12,13 In fact, reviewers for the 2012 Agency for Healthcare Research and Quality systematic review on the comparative effectiveness of treatments for osteoporosis excluded calcitonin “since most authorities no longer consider calcitonin to be appropriate treatment for osteoporosis.” ¹⁴

SCT is currently distributed in the United States by Novartis as Miacalcin and a recombinant form Fortical is distributed by Upsher Smith Pharmaceuticals.^4,15 Miacalcin is available in both an injectable and intranasal form at a 200 IU dosage. ⁴ Nasal Miacalcin has been approved for the treatment of PMO in the United States, whereas the injectable form has been approved for treatment of PMO, hypercalcemia, and Paget’s disease.^4,5 When SCT was first introduced, it was widely used, probably because hormone therapy was the only other PMO pharmaceutical available. In recent years, use of calcitonin for the treatment of PMO has declined substantially because of the availability of more efficacious drugs such as bisphosphonates. In 2011, calcitonin accounted for only 4.5% of all sales of PMO medications.^16,17 An oral formulation of SCT has been developed by Nordic Bioscience (in collaboration with Novartis) because it was a more acceptable route of administration for patients.^6,18 The oral formulation was thought to be efficacious in the treatment of both osteoarthritis and osteoporosis.^6,18-24 During the approval trials, a safety signal (concern about excess adverse events compared with level expected with a product’s use) for prostate cancer was observed, which prompted the review of all prior SCT trials regardless of the administration route to determine if use of SCT increases a patient’s risk for cancer.^16,17,25 To objectively understand the relationship between calcitonin and cancer and to judge how this relationship should influence SCT prescribing habits, we conducted an evidence review of the relevant literature and evaluated the strength of the data supporting the proposed association.

Data Sources and Selection

Study Selection and Data Extraction

Studies were included in our analysis if they were randomized, they had a follow-up period of ≥6 months, SCT was compared with a placebo, the full text was available in English, and rationale for early discontinuation or associated toxicities were collected. Studies were excluded if they did not include participants with cancer listed as an associated toxicity or reason for discontinuation, were using SCT for a malignant condition, were review articles, did not have a placebo control group, or combined SCT with another investigational agent or if their primary study goal was of either pharmacodynamic or pharmacokinetic nature. We reviewed 227 articles from MEDLINE/PubMed, 74 articles from MEDLINE/Ovid, 119 articles from EMBASE, and 21 articles (10 unpublished) from the Novartis and FDA briefing materials.^16,17 In all, 344 unique articles/studies were captured and reviewed using this method, with their collection and inclusion depicted in Figure 1. Information related to dosage, duration, cancer outcomes, relevance, and toxicities were collected by the study authors from these articles. In our analysis, we used the number of malignancy events provided to the FDA rather than the number of events in the published studies. Our search found 1 citation not specified by Novartis, which only differed by follow-up time, and 1 article that was not included in the Novartis meta-analysis.^39,42

OPEN IN VIEWER

Figure 1.

Flowchart of article selection.

We found 18 studies that collected any malignancy data; 15 had more events in the SCT arm, and 3 studies had more malignancy events in the placebo arm. Studies ranged from 12 to 60 months, with a median length of 24 months, and primarily included postmenopausal women (83%). Among them, 15 (83%) studies compared the nasal formulation with placebo, and 15 (83%) were likely initiated prior to 2000. Of the studies that had a malignancy event, 8 out of 18 (44%) were unpublished. A detailed review of the studies that met eligibility criteria is presented in Table 1. Information about the included studies has been collected from the published manuscripts, study reports from clinicaltrials.gov, the Novartis Briefing Book, and the FDA Background Document for the meeting of the Advisory Committee.^16,17 All cancer events recorded in the table are from the Novartis Briefing Book. ¹⁶ Discrepancies between sources are mentioned in the comments column of Table 1.

OPEN IN VIEWER

Table 1.

Detailed Information on the Studies That Collected Cancer Outcomes and Compared Salmon Calcitonin and Placebo.

Study Name	Year Initiated	Population	Route	Active Group (SCT Dose)	Other Therapies Given to Both Active and Placebo	Total Number Active From Published Manuscript	Cancer Events/Total Active Group (%) From Novartis Briefing Book 16	Cancer Events/Total Placebo Group (%) From Novartis Briefing Book 16	Attrition Rate (%) in Active/Placebo a	Length of Study	Citation	Comments
C2301 b	2007	OA	Oral	0.8 mg bid	None	588	22/585 (3.76)	6/584 (1.03)	33.0/22.8	24 Months	Clinicaltrials.gov 28	588 In both arms at start per clinicaltrials.gov
C2302 b	2008	OA	Oral	0.8 mg bid	None	521	21/520 (4.04)	14/508 (2.76)	42.4/33.4	24 Months	Clinicaltrials.gov 29	521 Active, 509 placebo at start per clinicaltrials.gov
A2303	2007	PMO	Oral	0.8 mg qd	800-1000 mg Calcium and 400-800 IU vitamin D	2334	89/2334 (3.81)	87/2331 (3.73)	32.4/25.7	36 Months	Clinicaltrials.gov 30
CT211	1989	PMO	Nasal	200 IU, 400 IU qd	None	200IU, 16; 400IU, 15	3/31 (9.68)	3/15 (20.0)	19/7/27	24 Months	Unpublished16,17
CT310	1988	PMO	Nasal	100 IU, 200 IU, 400 IU qd	None	100IU, 71; 200IU, 72; 400IU, 68	4/211 (1.90)	1/68 (1.47)	45/35/44/32	24 Months	Unpublished16,17
CT311 b	1992	PMO and GIO	Nasal	100 IU, 200 IU, 400 IU qd	None	244	8/244 (3.28)	4/79 (5.06)	45/35/44/32	24 Months (1 year open label FU)	Unpublished16,17
CT312 b	1992	PMO and GIO	Nasal	200 IU, 400 IU qd	None	201	11/201 (5.47)	3/102 (2.94)	39.2/33.3/33	24 Months (1 year open label FU)	Unpublished16,17
CT320	1991	PMO	Nasal	100 IU, 200 IU, 400 IU qd	1000 mg Calcium and 400 IU vitamin D qd	100IU, 316; 200IU, 316; 400IU, 312	81/944 (8.58)	16/311 (5.14)	60.8/58.2/59.3 /58.8	60 Months	Chesnut et al, 2000 8
SMCO522	Unk	PMO	Nasal	50 IU, 100 IU, 200 IU qd	500 mg Calcium qd	50IU, 52; 100IU, 52; 200IU, 52	4/156 (2.56)	0/52 (0.00)	21/12/15/15	24 Months	Overgaard et al, 1992 58
SMCO524	Unk	PMO	Nasal	100 IU, 200 IU, 400 IU qd	500 mg Calcium qd	100IU, 33; 200IU, 34; 400IU, 34	1/100 (1.00)	0/33 (0.00)	27.7/11.8/11.8 /12.1	24 Months	Overgaard, 1994 10	Total number active from published manuscript differs from Novartis total number
Unk	Unk	PMO	Nasal	100 IU qd	1000 mg Calcium qd	23	1/23 (4.35)	0/22 (0.00)	17.4/13.6	24 Months	Grigoriou et al, 1997 42
MIA16	1988	PMO c	Nasal	400 IU qd	1000 mg Calcium qd	32	1/32 (3.13)	0/30 (0.00)	19/10	24 Months (1 year open label FU)	Flicker et al, 1997 59	Calcium is recorded in published manuscript but not in Novartis Briefing Book
QUEST	Unk	PMO	Nasal	200 IU qd	500 mg Calcium carbonate qd	46	0/46 (0.00)	2/45 (4.44)	28/16	24 Months	Chesnut et al, 2005 45	Calcium is recorded in published manuscript but not in Novartis Briefing Book
SMCO005	1986	PMO with Colles’ fracture	Nasal	50 IU, 100 IU, 200 IU qd	500 mg Calcium carbonate qd	50 IU, 10; 100IU, 10; 200IU, 10	1/32 (3.13)	0/10 (0.00)	23.8% total	12 Months	Thamsborg et al, 1991 43	Total number active and attrition rate from published manuscript differ from Novartis values. Calcium is recorded in published manuscript but not in Novartis Briefing Book
SMCO504	1986	PMO	Nasal	100 IU qd	500 mg Calcium qd	29	1/29 (3.45)	0/29 (0.00)	41.4/37.9	24 Months	Unpublished16,17
SMCO506	1985	PMO	Nasal	50 IU 5× per week	500 mg Calcium 5× per week	142	3/147 (2.04)	1/141 (0.71)	31/33	36 Months	Reginster et al, 1994 44	Total number active and number placebo from published manuscript differ from Novartis Briefing Book
SMCO511	1987	Perimenopausal	Nasal	100 IU qd	1000 mg Calcium qd	60	2/60 (3.33)	0/60 (0.00)	32/37	36 Months	Arnala, 1996 39	SMCO511 specifications and results match Arnala’s study, but manuscript is not cited in the Novartis Briefing Book
SMCO514	Unk	PMO	Nasal	200 IU qd, 200 IU QMWF	None	200 IU qd, 36; 200 IU QMWF, 35	2/71 (2.82)	0/46 (0.00)	6/17/13	24 Months	Ellerington et al, 1996 11

Abbreviations: qd, once daily; bid, twice daily; QMWF, administered weekly on Monday, Wednesday, and Friday; PMO, postmenopausal osteoporosis; OA, osteoarthritis; GIO, glucocorticoid-induced osteoporosis; Unk, unknown; FU, follow-up.

a

Attrition rates from clinicaltrials.gov, reference 17 or published manuscript. If discrepancies with published manuscripts existed, data from reference 17 are shown.

b

Enrolled men.

c

Participants had prevalent fracture of spine, forearm, or femoral neck, except for 5 who had osteopenia at the lumbar spine without fracture.

Data Synthesis

Several issues were found with using the included studies to conclude a relationship between use of SCT and cancer. First, although types of cancer were individually analyzed, the main findings of the Novartis meta-analysis were based on general cancer risk. Using reports of any type of malignancy to make claims regarding the association could bias the results because of the differences in risk factors across the study populations.^11,43 Also, studies that reported cancer cases based on malignancy status after the trial, rather than excluding all prior cancer cases, could have led to an overestimation of cancer cases because recurrent cancer cases were counted as incident cases.^8,11 The failure to use standardized screening exams for cancer in these studies may have further caused those with prevalent/recurrent cancer to be enrolled and their cancer to be incorrectly attributed to use of the study drug. This is compounded by differing screening techniques depending on the country in which the trial was conducted and the time period wherein the study was conducted. ⁴⁴ With only 4 studies collecting outcomes for >2 years, these recurrent and undetected cancers may represent the majority of cancers included, with any cancer attributable to calcitonin use missed as a result of inadequate follow-up time.^8,44 Cancer information was collected systematically in only the recent SMC021 trials, with earlier trials possibly missing cancer cases because of potential nonblinding of SCT cases for interim bone density results or high attrition rates in either or both arms.^{8,10,16,17,28-30,42-44,45}

All studies used in both our analysis and the Novartis meta-analysis were RCTs. Although RCTs are considered the gold standard for testing associations, they are typically confined to very specific treatment populations, and their findings may or may not be generalizable to the general population using the drug.^46,47 The synthesis of these trials in a meta-analysis cannot fully take into account the differences in study protocol, populations, and outcomes to adequately assess an outcome that was not an a priori end point in any of the studies.^48,49 Furthermore, the lack of any biological, observational, or safety studies that showed an association between calcitonin and incidence of any specific type of cancer calls into question the use of pooled clinical trial data to appropriately assess the association.^16,17,33-36

The RCTs typically included dosages of SCT that are not approved for use worldwide, reducing the generalizability of the data. For example, the only nasal dosage of SCT approved in the United States is 200 IU. The Novartis meta-analysis did not find an association between SCT 200 IU daily and cancer (OR = 1.52; CI = 0.95-2.44) using a pooled sample size of 1528 individuals (nSCT 200 IU, 767; placebo, 761). ¹⁶ Thus, for countries that have only approved doses not associated with increased cancer, patients might incur no excess cancer risk with the use of SCT.

Information related to the trials with malignancy events was sparse and incomplete. Because 44% of these trials did not have published manuscripts and 83% were initiated prior to 2000, characterization of cancer cases was inhibited by a paucity of information relating to trial conduct and subject-level data and a significant amount of data that were not peer reviewed. Data from the unpublished reports had to be collected from the briefing materials and presentations to the FDA, with any data not published by either Novartis or the FDA missing in our analysis. Without more information on how these studies were conducted and their study populations and outcomes, properly weighting the applicability of their cancer cases is impossible and likely confounded the relationship found in the meta-analysis.^48,49

Discussion

We conducted an evidence review of a purported association between SCT use and cancer in postmenopausal women and older men. A safety signal and subsequent 2013 meta-analysis of SCT RCT data raised concern over a possible increased risk of cancer. After reviewing the same data and additional data from a literature search, we found the quality of the supporting evidence to be poor because of an overreliance on post hoc analysis, heterogeneity among trial methods and study populations, and insufficient longitudinal follow-up to assess cancer outcomes. The cancer event rates were very small, but a higher frequency of cancer occurred in the test groups compared with placebo groups among trials. Regardless of these findings, the overriding clinical shortcoming of SCT remains the absence of good-quality trial data supporting its ability to reduce fracture. In the 18 years since SCT’s approval for the treatment of PMO, multiple RCTs have demonstrated clear superiority of newer treatment agents to reduce fracture, with excellent safety profiles. We recommend that clinicians counsel their patients taking SCT for PMO to stop use now because the lack of evidence for fracture reduction alone justifies discontinuation. A possibility of cancer risk would only reinforce the decision to discontinue SCT use, so clinicians should not wait for the FDA’s decision to begin counseling patients.

Use of SCT for osteoporosis treatment has been controversial since the PROOF trial of nSCT versus placebo was published in 2000. ⁸ In this study, 59% of recruited study participants failed to complete the trial, leading to results that were difficult to interpret. Whereas BMD increases were dose dependent in PROOF, a dose-response effect for fracture reduction was not demonstrated. Thus, clinicians have continued to prescribe calcitonin based on surrogate end point (BMD) data, without good-quality RCT evidence of fracture reduction to date.

The strongest associations in the Novartis meta-analysis were for an aggregate general cancer risk, which increases with age even without the use of SCT. The lack of a significant increase for any specific cancer, other than basal cell carcinoma in postmenopausal women (primarily from the PROOF trial), may indicate that no excess cancer risk is present when SCT is used in the general population. We also find that the evidence of an increased cancer risk for SCT in men is limited because of the lack of events in the nasal arm of the C2301 and C2302 trials. Although the evidence for an increased risk for cancer with oSCT appears to be stronger, we do not feel that the association has been proven for both men and women or men alone. The Novartis meta-analysis of active versus placebo RCTs provides the greatest evidence of cancer risk, but with no reproducible basic research findings and no observational or retrospective studies of safety databases or observational cohorts, definitive causation cannot be proven. ⁵⁰ Although an RCT would more definitively determine if a risk exists, such a trial would be unethical and untimely. Before either the EMA or FDA label SCT as a known carcinogen and remove its indications for use, the limitations of the current RCTs and pooled meta-analysis need to be considered. The signal for cancer would ideally be studied in a prospective cohort study or high-quality population-based case-control study to help determine whether SCT use might be associated with an increase in new cancer cases or with promotion and progression of existing tumors. However, delaying a regulatory decision while waiting for new trial evidence would be unfeasible and potentially harmful.

Questions remain regarding the regulatory implications of the EMA’s and FDA’s decisions regarding SCT’s approval. If it is determined that the possible increased risk of cancer coupled with the reduced efficacy for fracture reduction are sufficient to remove the treatment indication for SCT, would this change in the FDA’s postmarketing approach be generalized to benefit-risk calculations for all drugs? If nSCT were to be released as a new drug in the United States today with only a possible increase in cancer risk, a risk evaluation and mitigation strategy likely would be established to determine if the risk exists with general population use. Should an approved drug be treated in the same manner, or should the only options be no monitoring or removal from the marketplace? Clarification of sound regulatory management of a drug’s risk-to-benefit ratio in the postmarket setting may be the most important result from the SCT safety signal. Resulting changes in regulatory policy would be consistent with the FDA’s goal of improved management of premarketing versus postmarketing uncertainties in benefits and risks of new agents and its handling of new findings that become available in the postmarket setting, where the basis for the finding comes from sources of varying levels of rigor. ⁵¹ However, before considering broader policy changes, the FDA would immediately need to review the approval of injectable SCT for the treatment of hypercalcemia and Paget’s disease to determine whether the risk of cancer might also outweigh potential benefits for these treatment indications. The latter decisions might be more complicated because no cancer cases from trials of injectable SCT were found and included in the Novartis meta-analysis.

To help guide clinical decision making, the US Preventive Services Task Force judges the clinical value of a preventive health care service by assessing the magnitude of benefits and harms, then estimating the net benefits to the patient. ⁵² Similarly, we need to weigh the potential benefits and risks of SCT use for fracture prevention while the FDA’s decision is pending. The ability of SCT to reduce vertebral fracture is questionable based on the PROOF trial data, and it has not been shown to reduce nonvertebral fractures. ⁵³ This weak efficacy contrasts with the significant reduction in vertebral, hip, and other nonvertebral fractures demonstrated in RCTs of bisphosphonates and denosumab.^54-56 Although SCT appears to reduce vertebral fracture pain, it has not been compared with NSAIDs or other pain medications in a head-to-head trial, and use of SCT solely for pain control is not substantiated. ⁵⁷ Considering SCT’s weak efficacy for fracture reduction and the current investigation of a possible association between SCT and cancer, we would encourage clinicians to choose more efficacious antiosteoporosis therapies for their patients unless contraindications exist.

We acknowledge that our search (and Novartis’ search for the Briefing Book presented to the FDA) was limited to RCTs. This was necessary to help ensure equivalent baseline cancer risk in SCT users versus nonusers and to avoid bias from concurrent or tandem use of other antifracture agents. Our review included studies found through MEDLINE and EMBASE searches. Because we identified only 1 additional study that was not included in the Novartis Briefing Book, we consider both our review and the data presented to the FDA panel to reflect the current literature.

We recommend that clinicians counsel their patients taking SCT for PMO to discontinue use now and to consider alternative agents that have demonstrated fracture reduction. The use of SCT is currently under review by both the European and United States pharmaceutical regulatory agencies for its possible association with an increased cancer risk. In our opinion, the evidence currently presented is not strong enough to prove a causal relationship between SCT and cancer. Regardless of the FDA’s and EMA’s decisions, SCT’s lack of demonstrated efficacy for fracture reduction alone justifies discontinuation, and a possibility of cancer risk leaves no sound rationale for extended therapy in elderly patients.